Transcriber’s Notes:

  Underscores “_” before and after a word or phrase indicate _italics_
    in the original text.
  Small capitals have been converted to SOLID capitals.
  Illustrations have been moved so they do not break up paragraphs.
  Typographical and punctuation errors have been silently corrected.




           Origin of Modern Calculating
                     Machines

        A chronicle of the evolution of the
         principles that form the generic
               make-up of the Modern
                Calculating Machine

                        BY
                  J. A. V. TURCK
    Member of The Western Society of Engineers

                   CHICAGO, 1921
      _Published under the auspices of_
         The Western Society of Engineers

                Copyright, 1921, by
                  J. A. V. Turck

[Illustration: Stone Age Calculating]




Foreword


There is nothing romantic in figures, and the average man takes little
interest in any subject pertaining to them. As a result of this
antipathy, there is plenty of historic evidence of man’s endeavor to
minimize the hated drudgery of calculation.

While history shows that, from prehistoric man down to the present age,
human ingenuity has turned to mechanical means to overcome the brain
fatigue of arithmetical figuring, it is within quite recent years that
he has really succeeded in devising means more rapid than the human
brain.

Of this modern product little has been written, except in disconnected
articles that have in no case offered a complete understanding as to
who were the great benefactors of mankind that gave to the world the
first concrete production of these modern principles of mechanical
calculation.

The writer, believing that there are many who would be interested to
know the true facts relative to this subject, has given to the public,
in that which follows, a chronicle of the evolution of the principles
disclosed in these modern machines, along with the proofs that form the
foundation for the story in a way that all may understand.

Although the subject has been handled in a way that makes it
unnecessary for the reader to be carried through a jangle of tiresome
mechanical construction, the writer believes that there are many
interested in the detail workings of these machines, and has for that
reason provided an interesting and simple description of the working
of each illustrated machine, which may be read by those who wish, or
skipped over, if the reader desires, without the danger of losing
knowledge of the relation of each of these machines to the Art.

[Illustration]




Chapters


                                                   PAGE
    Foreword                                         1
    Types of Ancient and Modern Machines             5
    The Early Key-Driven Art                        17
    The Key-Driven Calculator                       50
    Early Efforts in the Recording Machine Art      79
    First Practical Recorders                      111
    Introduction of the Modern Accounting Machine  144
    The High-Speed Calculator                      149
    The Improved Recorder                          163
    The Bookkeeping and Billing Machine            174
    A Closing Word                                 190




Illustrations


                                                                   PAGE
    Frontispiece, “Stone Age Calculating”
    One of the Pascal Machines                                      10
    Photo of Blaise Pascal                                          11
    Parmelee Patent Drawings                                        16
    Hill Patent Drawings                                            23
    Chapin Patent Drawings                                          28
    From the Stark Patent Drawings                                  32
    From the Robjohn Patent Drawings                                36
    From Drawings of Bouchet Patent 314,561                         40
    Drawings of Spalding Patent No. 293,809                         46
    “Macaroni Box” Model                                            53
    Photo of Dorr E. Felt                                           55
    The First “Comptometer”                                         57
    From Drawings of Felt Patent No. 371,496                        58
    Bill for First Manufacturing Tools of the Comptometer           68
    Early Comptometer                                               69
    Letter from Geo. W. Martin                                      71
    Testimonial                                                     72
    Testimonial                                                     73
    Letters from Elliott and Rosecrans                              74
    From Drawings of Barbour Patent No. 133,188                     78
    From Drawings of Baldwin Patent No. 159,244                     83
    Baldwin Machine                                                 83
    From Drawings of Pottin Patent No. 312,014                      88
    From Drawings of Burroughs Patent No. 388,118                   94
    Photo of Wm. S. Burroughs                                       95
    Drawings of Ludlum Patent No. 384,373                          104
    From Drawings of Felt Patent No. 405,024                       112
    Testimonial                                                    117
    Felt Recording and Listing Machine                             118
    From Drawings of Felt Patent No. 465,255                       121
    Felt Tabulator                                                 126
    One of the Early “Comptographs”                                130
    Photo of Gottfried Wilhelm Leibnitz                            132
    Leibnitz Calculator                                            133
    From Drawings of Burroughs’ Patents Nos. 504,963 and 505,078   136
    Burroughs’ Recorder                                            137
    From the February 1908 Issue of Office Appliances Magazine     142
    The High-Speed Calculator                                      148
    Two Pages from Wales Adding Machine Co. Booklet                165
    Moon-Hopkins Billing and Bookkeeping Machine                   176
    Napier’s Bones                                                 179
    From Drawings of Barbour Patent No. 130,404                    180
    Photo of John Napier                                           181
    From Drawings of Bollee Patent No. 556,720                     186




The Modern Accounting Machine


The term “adding machine” or “calculating machine” to most of us
represents the machine we have seen in the bank. The average person is
not familiar with the different types of accounting machines, to say
nothing of the many uses to which they are put; but he has a vague idea
that to hold any value they should produce a printed record, he doesn’t
know why and he hasn’t stopped to reason why; but those he has seen in
the bank do print, and any machine the bank uses, to his mind, must be
all right.

There are, of course, people who do know the different types of
accounting machines, and are familiar with their special uses, but
there are very few who are familiar with the true history of the modern
accounting machine.

[Sidenote: _General knowledge lacking_]

Articles written by those not familiar with the true facts relative to
the art of accounting machines have wrought confusion. Their errors
have been copied and new errors added, thus increasing the confusion.
Again, claims made in trade advertisements and booklets are misleading,
with the result that the truth is but little known.

These facts, and the psychological effect of seeing a certain type of
machine in the bank would lead the average man to believe that the
recording-adding machine was the only practical machine; and also (as
someone stated in the December, 1915, issue of the Geographic Magazine)
that Burroughs was the inventor of the recording-adding machine.

Although the history of accounting machines dates way back into
the tenth century, the modern accounting machines are of quite
recent origin, and are especially distinguished by the presence of
depressable keys. The keys in these machines act as a means of gauging
the actuation which determines the value in calculation, whether the
machine is key-driven or key-set with a crank or motor drive.

These modern machines, which come within the classification of
key-driven and key-set, have their respective special uses.

[Sidenote: _Key-driven machine first of the modern machines_]

The key-driven machine, which was the first produced of these two
types of modern machines, does not print, and is used for all forms of
calculation, but is generally behind the scenes in the accounting rooms
of all lines of business, and for that reason is not so well known as
the key-set crank-operated or motor-driven machine, which is designed
to print and is always in full view in the bank where it is used to
print your statement of account from the vouchers you have issued.

When we stop to analyze the qualities of these two types of machines,
we find that each has its place and that neither may truly serve to
displace the other. The organization of each is designed with reference
to the special work it was intended to do.

The calculating machine, having only to perform the work of
revolving the numeral wheels in calculating addition, subtraction,
multiplication and division in its many forms and combinations, may
be key-driven (on account of the slight mechanical resistance met with
in action), and thus, as a one-motion machine, requiring only the
depression of the keys, may also be much more rapid of manipulation
than the two-motion recording-adding machine which, after depressing
the keys for each item, requires the secondary operation of pulling a
crank forward or operating a push bar that connects the motor.

The recording-adding machine being designed to print the items and
answers of addition, requires power for the printing which cannot be
supplied by key depression. Thus an extra means for supplying that
power must be provided in the form of a crank lever, or in the latest
machines by a motor. The keys in such machines serve only as digital
control to gauge the setting of mechanism which prints the items and
adds them together. The secondary motion operates the mechanism to
print and add and finally to clear the machine for the setting up of
the next item. The recording of added columns of figures requires that
the answer must always be printed. This demands special operation of
devices provided for that purpose, which also adds to the time spent
in the operation of such machines as compared with the key-driven
calculator.

[Sidenote: _Recording, the primary feature of adding machines that
print_]

To state which of these two types of machines is the more useful would
cause a shower of comment, and has nothing to do with the object of
this article. Suffice it to say that where a printed record of items
added together with their answer is required for filing purposes,
or to bring together loose items like those in your bank statement,
the recording-adding machine serves; but when rapid calculation in
addition, multiplication, subtraction or division, or when combinations
of these forms of calculation are required, the key-driven calculator
is the practical machine for such work.

Although the key-driven calculator is generally not so well known, it
is, as stated, the oldest of the modern accounting machines, and its
usefulness places it in the accounting room, where it is oft-times
found employed by the hundreds in figuring up the day’s work of
accounting.

[Sidenote: _Validity and priority of invention_]

The purpose of this book is based wholly upon showing the validity and
priority of invention which constitute true contributions to the Art of
these two types of modern accounting machines; to place the facts for
once and all time before the public in such a way that they may judge
for themselves to whom the honor is due and thus settle the controversy
that exists.

The quibbling of court contests over the terminology of claims of
patents owned by the various inventors have been set aside and only
the true contributions to the Art which pertain to the fundamental
principles that have made the modern machines possible, are here dealt
with.

The dates of patents on inoperative or impractical machines have from
time to time been held up to the public as instances of priority of
invention; but when the validity of these patents, as furnishing any
real contributions to the Art, is questioned, they are not found to
hold the theme or principle that made the modern machines possible, and
as inventions, fade into obscurity.

[Illustration: _Figure 1_]

[Illustration: _Figure 2_ One of the Pascal Machines]

The Art of either the calculating machine or the adding-recording
machine is not new; it is, as a matter of fact, very old. As before
stated, the Art of “accounting machine” dates back to the tenth
century, but the first authentic evidence of a working machine is
extant in models made by Pascal in 1642 (see illustration).


THE PASCAL MACHINE

Referring to the illustration, Fig. 1, of Pascal’s machine on the
opposite page, it will be noted that there are a series of square
openings in the top of the casing; under these openings are drums, each
numbered on its cylindrical surface.

[Sidenote: _Description of Pascal’s invention_]

As the machine illustrated was made to figure English currency, the two
right-hand wheels are numbered for pence and shillings, while the six
wheels to the left are numbered from 1 to 9 and 0 for pounds.

[Illustration: Blaise Pascal]

The pounds register-drums, or numeral wheels, are each operated by a
train of gearing connecting them with a ten-armed turnstile wheel which
form the hub and spokes of what appears to be a series of wheels on the
top of the casing. While the spokes and hub are movable, the rims of
these wheels are stationary and are numbered from 1 to 9 and 0.

The geared relation between the turnstile wheels and the numeral wheels
is such that rotating a turnstile will give like rotation to its
numeral wheel.

Assuming that the numeral wheel of any one of the different orders
registered 0 through its sight opening and the turnstile of the same
order was moved one spoke of a rotation, it would move the wheel so
that the 0 would disappear and the figure 1 would appear; now if we
should move the same turnstile three more spokes the numeral wheel
would move likewise three spaces and the 4 would appear.

A stop in the form of a finger reaching over the spokes is provided
to stop the turnstile at the right point so that the figures on the
numeral wheels may register properly with the sight openings in the
casing.

[Sidenote: _Constructional features of the Pascal machine_]

The figures on the wheel rims fast to the casing are arranged
anti-clockwise to register with the space between the spokes, the 0
registering with the first space, the 1 with the second space and so
on around the wheel. Thus by use of the finger or a stylo inserted in
a space opposite the number to be added, the operator may move the
spoked wheel or turnstile clockwise until stopped by the stop finger.
By repeated selection and operation for each figure to be added, the
wheels will be revolved through their cycles of rotation caused by the
accumulation.

As the numeral wheels complete each rotation the 0 will appear, so
that a registration of the tens must be made. Pascal provided for the
accumulation of the tens by automatically turning the wheel of next
higher order one point through the action of the lower wheel.

The novel means employed for this transfer of the tens consisted of
a one-step ratchet device operated by a pin in the train of gearing
connected with the lower numeral wheel, which, as the lower wheel
passed from 9 to 0, forced the lever to which the ratchet pawl was
attached in a direction to cause the gearing of the higher numeral
wheel to be ratcheted forward far enough to add one to the higher
numeral wheel.

The direct actuation of a numbered wheel through its various degrees of
rotation and the secondary feature of effecting a one-step movement to
the numbered wheel of higher order (which seems to have been originated
by Pascal) is the foundation on which nearly all the calculating
machines have since been constructed to calculate the combinations
of the Arabian numerals represented in Addition, Multiplication,
Subtraction and Division.

In Fig. 2 of the illustration of Pascal’s machine, the machine has been
reversed, and the bottom of the casing, which is hinged, thrown back,
showing the numeral wheels and gearing of the different orders and the
transfer levers for the carry of the tens.

[Sidenote: _Increased capacity of modern calculator_]

The Art of the modern machines is far removed from the older Art by
its greatly increased capacity for rapid calculation which is found
emanating from the provision of keys as the means of manipulation.

To the unsophisticated, such a simple thing as applying keys to the
ancient type of calculating machines that have been made and used for
centuries, would seem but a simple mechanical application that the
ordinary mechanic could accomplish. But it was too great a problem for
the many renowned inventors of the older Art to solve.

Even though the use of depressable keys was common to many machines,
especially the piano, they knew that the organized make-up of their
machines could scarcely stand, without error, the slow action received
from the crank motion or other means employed as manipulating devices.
To place it within the power of an operator to operate their machines
at such a speed as would obtain in the sudden striking of a key would
result in chaos.

[Sidenote: _Patent office a repository of ineffectual efforts_]

There is no room for doubt that some of these early inventors had
the wish or desire to produce such a key-driven machine and may have
attempted to produce one. But as they lacked the advantage of an
institution like the Patent Office in which they could leave a record
of their inoperative inventions, and in view of the fact that they were
dependent on producing an operating machine for credit, there is no
authentic proof that they made attempts in this line.

[Illustration]

[Illustration: Parmelee Patent Drawings]




The Early Key-Driven Art


M. Le Colonel D’Ocagne, Ingénieur des Ponts et Chaussées, Professeur à
l’École des Ponts et Chaussées, Répétiteur à l’École Polytechnique, in
his “Le Calcul simplifie,” a historical review of calculating devices
and machines, refers to the key-driven machine as having first made its
appearance in the Schilt machine of 1851, but that the Art reached its
truly practical form in America. In the latter part of his statement
the professor is correct, but as to the first appearance of the
key-driven machine the U. S. Patent Office records show that a patent
was issued to D. D. Parmelee in 1850 for a key-driven adding machine
(see illustration).


THE PARMELEE MACHINE

[Sidenote: _First attempt to use depressable keys for adding was made
in America_]

By referring to the illustration of the Parmelee machine reproduced
from the drawings of the patent, the reader will notice that the
patentee deviated from the established principle of using numeral
wheels. In place of numeral wheels a long ratchet-toothed bar has been
supplied, the flat faces of which are numbered progressively from the
top to the bottom.

[Sidenote: _Description of Parmelee machine_]

As shown in Fig. 2 of these drawings, a spring-pressed ratchet pawl
marked k, engages the teeth of the ratchet or numeral bar. The pawl k,
is pivoted to a lever-constructed device marked E, the plan of which
is shown in Fig. 3. This lever device is pivoted and operated by the
keys which are provided with arms d, so arranged that when any one of
the keys is depressed the arm contacts with and operates the lever
device and its pawl k to ratchet the numeral bar upwards.

Another spring-pressed ratchet pawl marked m (see Fig. 2) is mounted
on the bottom of the casing and serves to hold the numeral bar from
returning after a key-depression.

It will be noted from Fig. 1 that the keys extend through the top of
the casing in progressively varying heights. This variation is such
as to allow the No. 1 key to ratchet up one tooth of the numeral bar,
the No. 2 key two teeth, etc., progressively. By this method a limited
column of digits could be added up by depressing the keys corresponding
to the digits and the answer could be read from the lowest tooth of the
numeral bar that protruded through the top of the casing.

It is evident that if the Parmelee machine was ever used to add with,
the operator would have to use a pussyfoot key-stroke or the numeral
bar would over-shoot and give an erroneous answer, as no provision was
made to overcome the momentum that could be given the numeral bar in an
adding action.

[Sidenote: _Foreign digit adders_]

[Sidenote: _Single digit adders lack capacity_]

The foreign machines of the key-driven type were made by V. Schilt,
1851; F. Arzberger, 1866; Stetner, 1882; Bagge, 1882; d’Azevedo, 1884;
Petetin, 1885; Maq Meyer, 1886. These foreign machines, like that of
Parmelee, according to M. le Colonel d’Ocagne, were limited to the
capacity of adding a single column of digits at a time. That is, either
a column of units or tens or hundreds, etc., at a time. Such machines,
of course, required the adding first of all the units, and a note made
of the total; then the machine must be cleared and the tens figure of
the total, and hundreds, if there be one, must then be added or carried
over to the tens column the same as adding single columns mentally.

On account of these machines having only a capacity for adding one
order or column of digits, the unit value 9 was the greatest item that
could be added at a time. Thus, if the overflow in adding the units
column or any other column amounted to more than one place, it required
a multiple of key-depressions to put it on the register. For example,
suppose the sum of adding the units columns should be 982, it would
require the depression of the 9-key ten times and then the 8-key to be
struck, to put the 98 on the machine. This order of manipulation had to
be repeated for each denominational column of figures.

Another method that could be used in the manipulation of these
single-order or digit-adding machines was to set down the sum of each
order as added with its units figure arranged relative to the order it
represents the sum of, and then mentally add such sums (see example
below) the same as you would set down the sums in multiplication and
add them together.

Example of method that may be used with single column adder.

        982
       563
      384
     125
    -------
     170012

Such machines, of course, never became popular because of their limited
capacity, which required many extra movements and caused mental strain
without offering an increase in speed of calculation as compared with
expert mental calculation. There were a number of patents issued in the
United States on machines of this class which may well be named single
digit adders.

[Sidenote: _Some early U.S. patents on single-digit adding machines_]

The machines of this type which were patented in the United States,
preceding the first practical multiple order modern machine, were
patented by D. D. Parmelee, 1850; W. Robjohn, 1872; D. Carroll, 1876;
Borland & Hoffman, 1878; M. Bouchet, 1883; A. Stetner, 1883; Spalding,
1884; L. M. Swem, 1885 and 1886; P. T. Lindholm, 1886; and B. F.
Smith, 1887. All of these machines varied in construction but not in
principle. Some were really operative and others inoperative, but all
lacked what may be termed useful capacity.

To those not familiar with the technical features of the key-driven
calculating machine Art, it would seem that if a machine could be made
to add one column of digits, it would require no great invention or
ingenuity to arrange such mechanisms in a plurality of orders. But the
impossibility of effecting such a combination without exercising a high
degree of invention will become evident as the reader becomes familiar
with the requirements, which are best illustrated through the errors
made by those who tried to produce such a machine.

As stated, the first authentic knowledge we have of an actual machine
for adding is extant in models made by Pascal in 1642, which were all
multiple-order machines, and the same in general as that shown in the
illustration, page 10.

[Sidenote: _Calculating machines in use abroad for centuries_]

History shows that Europe and other foreign countries have been using
calculating machines for centuries. Like that of Pascal’s, they were
all multiple-order machines, and, although not key-driven, they were
capable of adding a number of columns or items of six to eight places
at once without the extra manipulation described as necessary with
single-order digit adding machines. A number of such machines were
made in the United States prior to the first practical multiple-order
key-driven calculator.

[Sidenote: _First key-driven machines no improvement to the Art_]

This fact and the fact that the only operative key-driven machines
made prior to 1887 were single-digit adders are significant proof that
the backward step from such multiple-order machines to a single-order
key-driven machine was from the lack of some unknown mechanical
functions that would make a multiple-order key-driven calculator
possible. There was a reason, and a good one, that kept the inventors
of these single-order key-driven machines from turning their invention
into a multiple-order key-driven machine.

It is folly to think that all these inventors never had the thought or
wish to produce such a machine. It is more reasonable to believe there
was not one of them who did not have the wish and who did not give deep
thought to the subject. There is every reason to believe that some of
them tried it, but there is no doubt that if they did it was a failure,
or there would be evidence of it in some form.


THE HILL MACHINE

The U. S. Patent Office records show that one ambitious inventor,
Thomas Hill, in 1857 secured a patent on a multiple-order key-driven
calculating machine (see illustration), which he claimed as a new and
useful invention. The Hill patent, however, was the only one of that
class issued, until the first really operative modern machine was made
thirty years later, and affords a fine example by which the features
that were lacking in the make-up of a really operative machine of this
type may be brought out.

[Sidenote: _Description of the Hill machine_]

The illustrations of the Hill machine on the opposite page, reproduced
from the drawings of the patent, show two numeral wheels, each having
seven sets each of large and small figures running from 1 to 9 and
the cipher marked on their periphery. The large sets of figures are
arranged for addition or positive calculation, and the small figures
are arranged the reverse for subtraction or negative calculation. The
wheels are provided with means for the carry of the tens, very similar
to that found in the Pascal machine. Each of the two wheels shown are
provided with ratchet teeth which correspond in number with the number
of figures on the wheel.

Spring-pressed, hook-shaped ratchet pawls marked b, are arranged to be
in constant engagement with the numeral wheels. These pawls are each
pivotally mounted in the end of the levers marked E, which are pivoted
at the front end of the casing.

[Illustration: Hill Patent Drawings]

The levers E, are held in normal or upward position by springs f, at
the front of the machine. Above each of these levers E, are a series of
keys which protrude through the casing with their lower ends resting
on the levers. There are but six keys shown in the drawing, but the
specification claims that a complete set of nine keys may be supplied
for each lever.

The arrangement and spacing of the keys are such that the greater the
value of the key the nearer it is to the fulcrum or pivot of the lever
E. The length of the key stem under the head or button of each key is
gauged to allow depression of the key, the lever E and pawl b, far
enough to cause the numeral wheel to rotate as many numeral places as
the value marking on the key.

A back-stop pawl for the numeral wheels, marked p, is mounted on a
cross-rod at the top of the machine. But one of these pawls are shown,
the shaft and the pawl for the higher wheel being broken away to show
the device for transferring the tens to the higher wheel.

The transfer device for the carry of the tens is a lever arrangement
constructed from a tube F, mounted on the cross-rod m, with arms G and
H. Pivoted to the arm G, is a ratchet pawl i, and attached to the pawl
is a spring that serves to hold the pawl in engagement with the ratchet
of the higher-order numeral wheel, and at the same time, through its
attachment with the pawl, holds the lever arms G and H retracted as
shown in the drawing.

As the lower-order numeral wheel passes any one of its points from 9
to O, one of the teeth or cam lugs n, on the wheel will move the arm
H, of the transfer lever forward, causing the pawl i, to move the
higher-order wheel one step to register the accumulation of the tens.

The functions of the Hill mechanism would, perhaps, be practical if it
were not for the physical law that “bodies set in motion tend to remain
in motion.”

[Sidenote: _Hill machine at National Museum_]

Considerable unearned publicity has been given the Hill invention on
account of the patent office model having been placed on exhibit in the
National Museum at Washington. Judging from the outward appearance of
this model, the arrangement of the keys in columns would seem to impart
the impression that here was the foundation of the modern key-driven
machine. The columnar principle used in the arrangement of the keys,
however, is the only similarity.

[Sidenote: _Inoperativeness of Hill machine_]

The Hill invention, moreover, was lacking in the essential feature
necessary to the make-up of such a machine, a lack that for thirty
years held the ancient Art against the inroads of the modern Art that
finally displaced it. The feature lacking was a means for controlling
the action of the mechanism under the tremendously increased speed
produced by the use of depressable keys as an actuating means.

Hill made no provision for overcoming the lightning-speed momentum that
could be given the numeral wheels in his machine through manipulation
of the keys, either from direct key-action or indirectly through the
carry of the tens. Imagine the sudden whirl his numeral wheel would
receive on a quick depression of a key and then consider that he
provided no means for stopping these wheels; it is obvious that a
correct result could not be obtained by the use of such mechanism. Some
idea of what would take place in the Hill machine under manipulation
by an operator may be conceived from the speed attained in the
operation of the keys of the up-to-date modern key-driven machine.

[Sidenote: _High speed of key drive_]

Operators on key-driven machines oftentimes attain a speed of 550 key
strokes a minute in multiplication. Let us presume that any one of
these strokes may be a depression of a nine key. The depression and
return, of course, represents a full stroke, but only half of the
stroke would represent the time in which the wheel acts. Thus the
numeral wheel would be turned nine of its ten points of rotation in an
eleven hundredth (¹/₁₁₀₀) of a minute. That means only one-ninth of
the time given to half of the key-stroke, or a ninety-nine hundredth
(¹/₉₉₀₀) of a minute; a one hundred and sixty-fifth (¹/₁₆₅) part of a
second for a carry to be effected.

[Sidenote: _Camera slow compared with carry of the tens_]

If you have ever watched a camera-shutter work on a twenty-fifth of
a second exposure, which is the average time for a snap-shot with an
ordinary camera, it will be interesting to know that these controlling
devices of a key-driven machine must act in one-fifth the time in which
the shutter allows the daylight to pass through the lens of the camera.

Think of it; a machine built with the idea of offering the possibility
of such key manipulation and supplying nothing to overcome the
tremendous momentum set up in the numeral wheels and their driving
mechanism, unless perchance Hill thought the operator of his machine
could, mentally, control the wheels against over-rotation.

[Illustration]

[Illustration: Chapin Patent Drawings]

Lack of a proper descriptive term used to refer to an object, machine,
etc., oftentimes leads to the use of an erroneous term. To call the
Hill invention an adding machine is erroneous since it would not add
correctly. It is as great an error as it would be to refer to the
Langley aeroplane as a flying machine.

[Sidenote: _Hill machine merely adding mechanism, incomplete as
operative machine_]

When the Wright brothers added the element that was lacking in the
Langley plane, a real flying machine was produced. But without that
element the Langley plane was not a flying machine. Likewise, without
means for controlling the numeral wheels, the Hill invention was not an
adding machine. The only term that may be correctly applied to the Hill
invention is “adding mechanism,” which is broad enough to cover its
incompleteness. And yet many thousands of people who have seen the Hill
invention at the National Museum have probably carried away the idea
that the Hill invention was a perfectly good key-driven adding machine.

[Sidenote: _Chapin and Stark patents_]

Lest we leave unmentioned two machines that might be misconstrued to
hold some of the features of the Art, attention is called to patents
issued to G. W. Chapin in 1870 (see illustration on opposite page), and
A. Stark in 1884 (see illustration on page 32).


CHAPIN MACHINE

[Sidenote: _Description of Chapin machine_]

Referring to the illustration reproducing the drawings of the Chapin
patent, the reader will note that in Fig. 1 there are four wheels
marked V. These wheels, although showing no numerals, are, according to
the specification, the numeral wheels of the machine.

The wheels are provided with a one-step ratchet device for transferring
the tens, consisting of the spring frame and pawl shown in Fig. 3,
which is operated by a pin in the lower wheel.

In Fig. 1 the units and tens wheel are shown meshed with their driving
gears. These gears are not numbered but are said to be fast to the
shafts N and M, respectively (see Fig. 2).

Fast on the shaft M, is a series of nine ratchet-toothed gears marked
O, and a like series of gears P, are fast to the shaft N. Co-acting
with each of these ratchet-toothed gears is a ratchet-toothed rack F,
pivoted at its lower end to a key-lever H, and pressed forward into
engagement with its ratchet gear by a spring G.

The key-levers H, of which there are two sets, one set with the
finger-pieces K and the other with the finger-pieces J, are all pivoted
on the block I, and held depressed at the rear by an elastic band L.
The two sets of racks F, are each provided with a number of teeth
arranged progressively from one to nine, the rack connected with the
No. 1 key having one ratchet tooth, the No. 2 having two teeth, etc.

[Sidenote: _Inoperativeness of Chapin machine_]

By this arrangement Chapin expected to add the units and tens of a
column of numerical items, and then by shifting the numeral wheels and
their transfer devices, which are mounted on a frame, designed for that
purpose, he expected to add up the hundred and thousands of the same
column of items.

It is hardly conceivable that the inventor should have overlooked the
necessity of gauging the throw of the racks F, but such is the fact, as
no provision is made in the drawings, neither was mention made of such
means in the specification. Even a single tooth on his rack F, could,
under a quick key-stroke, overthrow the numeral wheels, and the same is
true of the carry transfer mechanism.

[Illustration]

[Illustration: From the Stark Patent Drawings]

The Chapin machine, like that of Hill, was made without thought as
to what would happen when a key was depressed with a quick stroke,
as there was no provision for control of the numeral wheels against
overthrow. As stated, the machine was designed to add two columns
of digits at a time, and with an attempt to provide means to shift
the accumulator mechanism, or the numeral wheels and carry-transfer
devices, so that columns of items having four places could be added
by such a shift. Such a machine, of course, offered less than could
be found in the Hill machine, and that was nothing at all so far as a
possible operative machine is concerned.


THE STARK MACHINE

The reproduction of the patent drawings of the Stark machine
illustrated on the opposite page show a series of numeral wheels, each
provided with three sets of figures running from 1 to 9 and 0.

[Sidenote: _Description of Stark machine_]

Pivotally mounted upon the axis of the numeral wheels at each end
are sector gears E¹ and arms E⁴, in which are pivoted a square shaft
E, extended from one arm to the other across the face of the numeral
wheels. The shaft E, is claimed to be held in its normal position by a
spring so that a pawl, E², shiftably mounted on the shaft, designed to
ratchet or actuate the numeral wheels forward, may engage with any one
of the numeral wheel ratchets.

A bail marked D, is pivoted to standards A¹, of the frame of the
machine, and is provided with the two radial racks D³ which mesh with
the sector gears E¹. It may be conceived that the act of depressing
the bail D, will cause the actuating pawl E², to operate whichever
numeral wheel it engages the ratchet of.

The bail D, is held in its normal position by a spring D², and is
provided with nine keys or finger-pieces d, eight of which co-act with
the stepped plate G, to regulate the additive degree of rotation given
to the numeral wheels, while the ninth has a fixed relation with the
bail and the bail itself is stopped.

The keys d, marked from 1 to 8, are pivoted to the bail in such a
manner that their normal relation to the bail will allow them to pass
by the steps on the stepped plate G, when the bail is depressed by
the fixed No. 9 key. When, however, any one of the keys numbered from
1 to 8 is depressed, the lower end of the shank of the key will tilt
rearward, and, as the bail is depressed, offers a stop against the
respective step of the plate G, arranged in its path, thus stopping
further action of the actuating pawl E², but offering nothing to
prevent the continuation of the force of momentum set up in the numeral
wheels by the key action.

There was small use in stopping the action of the pawl E², if the
ratchet and numeral wheel, impelled by the pawl, could continue onward
under its momentum.

The carry of the tens transfer device is of the same order as that
described in the Pascal and Hill machines; that is, a one-step
ratchet-motion actuated by a cam lug or pin from the lower wheel. The
carry transfer device consists of the lever F, and pawl f⁴, acting on
the ratchet of the upper wheel which is operated by the cam lugs b⁵ of
the lower wheel acting on the arms f¹ and f³ of the lever F.

[Illustration: From the Robjohn Patent Drawings]

[Sidenote: _Inoperativeness of Stark machine_]

The machine shown in the Stark patent was provided with but one set of
keys, but the arrangement for shifting the driving ratchet pawl E²,
from one order to another, so that the action of the keys may rotate
any one of the numeral wheels, gave the machine greater capacity than
the single digit adders; but as with the Chapin machine, of what use
was the increase in capacity if the machine would not add correctly.
That is about all that may be said of the Stark machine, for since
there was no means provided by which the rotation of numeral wheels
could be controlled, it was merely a device for rotating numeral wheels
and was therefore lacking in the features that would give it a right to
the title of an adding machine.

[Sidenote: _Nine keys common to a plurality of orders_]

The nine-key scheme of the Stark invention, connectable to the
different orders, was old, and was first disclosed in the U. S. Patent
to O. L. Castle in 1857 (a machine operated by a clock-spring wound by
hand), but its use in either of these machines should not be construed
as holding anything in common with that found in some of the modern
recording adders. The Castle machine has not been illustrated because
it does not enter into the evolution of the modern machine.

The ancient Art, or the Art prior to the invention of Parmelee,
consisted of mechanism which could be controlled by friction devices,
or Geneva gear-lock devices, that were suitable to the slow-acting type
of manipulative means.

The first attempt at a positive control for a key-driven adding device
is found in a patent issued to W. Robjohn in 1872 (see illustration).
As will be noted, this machine was referred to in the foregoing
discussion as merely a single-digit adding machine, having the capacity
for adding but one column of digits at a time.


ROBJOHN MACHINE

Referring to the illustration of the patent drawings of the Robjohn
machine, it will be noted that there are three sight openings in the
casing through which the registration of the numeral wheels may be
read. The numeral wheels, like those of all machines of this character,
are connected by devices of a similar nature to those in the Hill
machine for carrying the tens, one operating between the units and tens
wheel and another between the tens and hundredths wheel.

[Sidenote: _Description of Robjohn machine_]

The units wheel shown in Fig. 3 is connected by gearing to a long
pin-wheel rotor, marked E, so that any rotation of the rotor E,
will give a like rotation to the units numeral wheel to which it is
entrained by gearing.

To each of the nine digital keys, marked B, is attached an engaging and
disengaging sector gear device, which, as shown in Fig. 3, although
normally not in engagement with the rotor E, will upon depression of
its attached key, engage the rotor and turn it.

A stop device is supplied for the key action, which in turn was
supposed to stop the gear action; that seems rather doubtful. However,
an alternative device is shown in Figs. 4 and 5, which provides what
may without question be called a stop device to prevent over-rotation
of the units wheel under direct key action.

[Illustration: From Drawings of Bouchet Patent 314,561]

It will be noted that the engaging and disengaging gear device is here
shown in the form of a gear-toothed rack and that the key stem is
provided with a projecting arm ending in a downwardly projecting tooth
or detent which may engage the rotor E, and stop it at the end of the
downward key action. While the stopping of the rotor shows a control in
the Robjohn machine which takes place under direct action from the keys
to prevent overthrow of the units numeral wheel, it did not prevent the
overflow of the higher or tens wheels, if a carry should take place.
There was no provision for a control of the numeral wheels under the
action received from the carry of the tens by the transfer mechanism.

[Sidenote: _First control for a carried numeral wheel_]

The first attempt to control the carried wheel in a key-driven machine
is found in a patent issued to Bouchet in 1882 (see illustration
on opposite page); but it was a Geneva motion gearing which, as is
generally known, may act to transmit power and then act to lock the
wheel to which the power has been transmitted until it is again to be
turned through the same source. Such a geared up and locked relation
between the numeral wheels, of course, made the turning of the higher
wheel (which had been so locked) by another set of key-mechanism an
impossibility.


BOUCHET MACHINE

The illustration of the Bouchet machine on the opposite page was
reproduced from the drawings of the patent which is the nearest to the
machine that was placed on the market. The numeral wheels, like most of
the single-digit adders, are three in number, and consist of the prime
actuated, or units wheel, and two overflow wheels to receive the carry
of the tens. The units wheel has fixed to it a long 10-tooth pinion or
rotor I, with which nine internal segmental gear racks L, are arranged
to engage and turn the units wheel through their nine varying additive
degrees of rotation.

[Sidenote: _Description of Bouchet machine_]

The segmental gear racks L, are normally out of mesh with the pinion
I, and are fast to the key levers E, in such a manner that the first
depression of a key causes its rack to rock forward and engage with the
pinion I, and further depression moves the rack upward and rotates the
pinion and units numeral wheel. It will be noted that this engaging and
disengaging gear action is in principle like that of Robjohn.

The transfer devices for the carry of the tens, as already stated,
belong to that class of mechanism commonly known as the “Geneva
motion.” It consists of a mutilated or one-tooth gear fast to the units
wheel operating with a nine-tooth gear, marked D¹, loosely mounted on
an axis parallel to the numeral wheel axis. Each revolution of the
units wheel moves the nine-tooth gear three spaces, and in turn moves
the next higher numeral wheel to which it is geared far enough to
register one point or the carry. A circular notched disc, marked S, is
fast to the units wheel, and the nine-tooth gear D¹, has part of two
out of every three of its teeth mutilated or cut away to make a convex
surface for the notched disc to rotate in.

With such construction the nine-tooth gear may not rotate or become
displaced as long as the periphery of the disc continues to occupy
any one of the three convex spaces of the nine-tooth gear. When,
however, the notch of the disc is presented to the mutilated portion
of the nine-tooth gear, the said gear is unlocked. This unlocking
is coincident to the engagement of the single tooth of the numeral
wheel-gear with the nine-tooth gear and the passing of the numeral
wheel from 9 to 0, during which the nine-tooth gear will be moved three
spaces, and will be again locked as the notch in the disc passes and
the periphery fills the next convex space of the mutilated nine-tooth
gear.

[Sidenote: _Bouchet machine marketed_]

The Bouchet machine was manufactured and sold to some extent, but
never became popular, as it lacked capacity. Machines of such limited
capacity could not compete with ordinary accountants, much less with
those who could mentally add from two to four columns at a clip.
Aside from the capacity feature, there was another reason why these
single-order machines were useless, except to those who could not
add mentally. Multiple forms of calculation, that is, multiplication
and division, call for a machine having a multiplicity of orders.
The capacity of a single order would be but 9 × 9, which requires no
machine at all--a seven-year-old child knows that. To multiply 58964
× 6824, however, is a different thing, and requires a multiple-order
calculator.

[Sidenote: _Misuse of the term “Calculating Machine”_]

It is perhaps well at this time to point out the misuse of the term
calculating where it is applied to machines having only a capacity
for certain forms of calculating as compared with machines which
perform in a practical way all forms of calculation, that is,
addition, multiplication, subtraction and division. To apply the term
“calculating machine” to a machine having anything less than a capacity
for all these forms is erroneous.

An adding machine may perform one of the forms of calculation, but to
call it a calculating machine when it has no capacity for division,
subtraction or multiplication, is an error; and yet we find the U. S.
Patent Office records stuffed full of patents granted on machines thus
erroneously named. The term calculating is the broad term covering all
forms of calculation, and machines performing less should be designated
according to their specific capacities.

It is true that adding is calculating, and under these circumstances,
why then may not an adding machine be called a calculator? The answer
is that it may be calculating to add; it may be calculating to either
subtract, multiply or divide; but if a machine adds and is lacking in
the means of performing the other forms of calculation, it is only part
of a calculating machine and lacks the features that will give it title
to being a full-fledged calculator.[1]

[1] NOTE: The title of this book does not coincide with the above
argument, but in view of the common use of the term “calculating” its
application is better understood.

Considerable contention was raised by parties in a late patent suit as
to what constituted the make-up of a calculating machine. One of the
attorneys contended that construction was the only thing that would
distinguish a calculating machine. But as machines are named by their
functioning, the contention does not hold water. That is to say: A
machine may be a calculating machine and yet its construction be such
that it performs its functions of negative and positive calculation
without reversal of its action.

Again, a machine may be a calculating machine and operate in one
direction for positive calculation and the reverse for negative
calculation. As long as the machine has been so arranged that all forms
of calculation may be performed by it without mental computation, and
the machine has a reasonable capacity of at least eight orders, it
should be entitled to be called a calculating machine.

[Illustration: Drawings of Spalding Patent No. 293,809]


THE SPALDING MACHINE

The next machine that has any bearing on the key-driven Art of which
there is a record, is illustrated in a patent granted to C. G. Spalding
in 1884 (see illustration on opposite page). The Spalding invention,
like that of Bouchet, was provided with control for its primary
actuation and control for its secondary or carrying actuation.

[Sidenote: _Description of Spalding machine_]

Referring to the Spalding machine reproduced from the drawings of
his patent, the reader will note that in place of the units and tens
numeral wheels, a clock hand has been supplied, co-operating with a
dial graduated from 0 to 99, showing the figures 5, 10, 15, etc., to
95, for every five graduations.

Another similar hand or arrow and dial to register the hundreds is also
provided, having a capacity to register nineteen hundred. Attached to
the arrows, through a shaft connection at the back of the casing are
ratchet wheels, having respectively the same number of teeth as the
graduation of the dial to which each hand belongs.

Co-operating with the hundred-tooth ratchet of the units and tens
register hand is a ratchet and lever motion device (see Fig. 2) to turn
the arrow from one to nine points of the graduation of the dial. The
ratchet and lever motion device consists of the spring-pressed pawl E,
mounted on the lever arm D, engaging the hundred-tooth ratchet, the
link or push-rod F, the lever G, and its spring O. It will be noted
that a downward action of the lever G, will, through the rod F, cause
a like downward action of the lever D, causing the ratchet pawl E to
be drawn over the ratchet teeth. Upon the release of the lever G, the
spring O, will return it to its normal position and through the named
connecting parts, ratchet forward the arrow.

The normal position of the pawl E is jammed into the tooth of the
ratchet and against the bracket C, that forms the pivot support for
the pivot shaft of the arrow. This jammed or locked combination serves
to stop the momentum of the ratchet wheel at the end of the ratcheting
action, and holds the wheel and its arrow normally locked until the
lever G is again depressed.

The means for gauging the depression and additive degrees of action of
the lever G is produced through the slides or keys marked a, having
finger-pieces c, springs f, and pins e, bearing against the top of the
lever G, combined with what may be called a compensating lever marked K.

The specification of the patent states that the depression of a key
will depress the lever G and the free end will engage the bent end t,
of the compensating lever K, and rock its envolute curved arm M, upward
until it engages the pin e of the key, which will block further motion
of the parts.

The effectiveness of the construction shown for the lever K is open to
question.

[Sidenote: _Prime actuation of a carried wheel impossible in the
Spalding machine_]

The carry of the hundreds is accomplished by means of a one-step
ratchet device represented by the parts lever R, pawl T, spring P, and
operating pin g. When the hundred-tooth ratchet nears the end of its
revolution, the pin g, made fast therein, engages the free end of the
ratchet lever R, and depresses it; and as the hand attached to the
hundred-tooth ratchet wheel passes from 99 to 0 the pin g passes off
the end of the ratchet lever R, and the spring P retracts the lever
ratcheting the twenty-tooth wheel and its arrow forward one point so
that the arrow registers one point greater on the hundreds dial.

Although the Spalding means of control under carrying differed from
that of Bouchet in construction, its function was virtually the same
in that it locked the carried or higher wheel in such a manner as
to prevent the wheel from being operated by an ordinal set of key
mechanism.

And the control under key action would prevent a carry being delivered
to that order through the locked relation of the ratchet and pawl E.

[Illustration]




The Key-Driven Calculator


While these single-digit adding machines have been used to illustrate
how the control, which was lacking in the Hill invention, had been
recognized by other inventors as a necessary requisite to the
key-drive, it should not be construed that such carrying control as had
been applied to their inventions was of a type that could be used in
the Hill machine or in any multiple-order key-driven machine. It was
thirty years after the first attempt to control a key-driven machine
was made before an operative multiple-order key-driven machine, with a
control that would prevent over-rotation, was finally invented.

[Sidenote: _Theory versus the concrete_]

Theoretically, it would seem that the only feature or element lacking
in the Art prior to 1886, to produce a real key-driven calculator was
means that would control the carrying and also leave the carried wheel
free for key actuation. It was, however, quite a different problem.
Theoretical functions may be patched together to make a theoretical
machine; but that is only theory and not the concrete.

[Sidenote: _All but one of the generic elements solved_]

To take fragmental parts of such machines as were disclosed in the Art
and patch them together into anything practical was impossible, even
if one had been familiar with the Art and could devise mechanism to
supply the new element. That is, leaving aside the broad or generic
theoretical elements, which today, from knowledge gained by later
inventions, serve the make-up of a key-driven calculator, there was
still lacking any concrete example or specific design of a whole
machine, as there was no such machine disclosed in the drawings of
patents, or any known mechanism which, if arranged in multiples, would
be operative as a practical machine even if mechanism to supply the new
element were to be added.

In other words, while it is conceded from our present knowledge that
all but one of the generic theoretical elements had been solved as
disclosed in the various before-named machines, it required the
application of these elements in a different way from anything before
disclosed; which in itself required a different concrete form of the
generic principles for the whole machine as well as a generic form of
invention covering the new theoretical element.

It may be easy to analyze that which exists, but quite a different
story to conceive that which did not exist. With reference to the Art,
however, the production of the new element is a feature that may be
credited without question. The concrete does not enter into it other
than as proof that a new feature has been created.

[Sidenote: _Originality of inventions_]

While the discussion of the Art from a scientific standpoint brings
together in after years what has been accomplished by different
inventors, it is doubtful whether any of these early inventors had
other knowledge than what may possibly have been obtained from seeing
one of the foreign-made crank-driven machines. All inventors work with
an idea obtained from some source, but on the whole few copy inventions
of others. When an Art is fully established, however, and machines
representing the Art are to be found on the market and the principal
features of such machines are portrayed in a later patent, it may
rightly be called a copy. To assume, however, that a novice has taken
the trouble to delve into the archives of the patent office and study
the scattered theoretical elements of the Art and supply a new element
to make a combination that is needed to produce a practical key-driven
calculator, is not a probable assumption. But allowing such assumption
were possible, it is evident that from anything that the Art disclosed
prior to 1887 it was not possible to solve the concrete production of a
key-driven calculator.

[Sidenote: _A conception which led to the final solution_]

In 1884, a young machinist, while running a planer, conceived an idea
from watching its ratchet feed motion, which was indirectly responsible
for the final solution of the multiple-order key-driven calculating
machine. The motion, which was like that to be found on all planing
machines, could be adjusted to ratchet one, two, three, four or more
teeth for a fine or coarse feed.

While there is nothing in such a motion that would in any way solve the
problem of the modern calculator, it was enough to excite the ambitions
of the man who did finally solve it. It is stated that the young man,
after months of thought, made a wooden model, which he finished early
in 1885. This model is extant, and is illustrated on the opposite page.

The inventor was Dorr E. Felt, who is well known in the calculating
machine Art as the manufacturer of the “Comptometer,” and in public
life as a keen student of economic and scientific subjects. The wooden
model, as will be noted, was crude, but it held the nucleus of the
machine to come.

[Illustration: “Macaroni Box” Model]

[Illustration: Dorr E. Felt]

Mr. Felt has given some interesting facts regarding his experience in
making the wooden model.

[Sidenote: _Evolution of an invention_]

He says: “Watching the planer-feed set me to scheming on ideas for
a machine to simplify the hard grind of the bookkeeper in his day’s
calculation of accounts.

“I realized that for a machine to hold any value to an accountant, it
must have greater capacity than the average expert accountant. Now I
knew that many accountants could mentally add four columns of figures
at a time, so I decided that I must beat that in designing my machine.
Therefore, I worked on the principle of duplicate denominational
orders that could be stretched to any capacity within reason. The plan
I finally settled on is displayed in what is generally known as the
“Macaroni Box” model. This crude model was made under rather adverse
circumstances.

“The construction of such a complicated machine from metal, as I had
schemed up, was not within my reach from a monetary standpoint, so I
decided to put my ideas into wood.

[Sidenote: _Trials of an inventor_]

“It was near Thanksgiving Day of 1884, and I decided to use the holiday
in the construction of the wooden model. I went to the grocer’s and
selected a box which seemed to me to be about the right size for the
casing. It was a macaroni box, so I have always called it the macaroni
box model. For keys I procured some meat skewers from the butcher
around the corner and some staples from a hardware store for the key
guides and an assortment of elastic bands to be used for springs. When
Thanksgiving day came I got up early and went to work with a few tools,
principally a jack knife.

“I soon discovered that there were some parts which would require
better tools than I had at hand for the purpose, and when night came I
found that the model I had expected to construct in a day was a long
way from being complete or in working order. I finally had some of the
parts made out of metal, and finished the model soon after New Year’s
day, 1885.”

[Sidenote: _The first “Comptometer”_]

By further experimenting the scheme of the wooden model was improved
upon, and Felt produced, in the fall of 1886, a finished practical
machine made of metal. This machine is illustrated on the opposite page.


THE FELT CALCULATING MACHINE

Referring to the illustration of Felt’s first metal machine, it will be
noted that the machine has been partly dismantled. The model was robbed
of some of its parts to be used as samples for the manufacture of a
lot of machines that were made later. In view of the fact that this
machine is the first operative multiple-order key-driven calculating
machine made, it seems a shame that it had to be so dismantled; but the
remaining orders are operative and serve well to demonstrate the claims
held for it.

[Sidenote: _Felt patent 371,496_]

The mechanism of the machine is illustrated in the reproduction of the
drawings of Felt’s patent, 371,496, on page 58. The specification of
this patent shows that it was applied for in March, 1887, and issued
October 11, 1887.

From the outward appearance of the machine it has the same general
scheme of formation as is disclosed in the wooden model.

[Illustration: The First “Comptometer”]

[Illustration]

[Illustration: From Drawings of Felt Patent No. 371,496]

[Sidenote: _Description of Felt calculator_]

The constructional scheme of the mechanism consists of a series of
numeral wheels, marked A in the patent drawings. Each wheel is provided
with a ratchet wheel, and co-acting with the ratchet is a pawl mounted
on a disc E², carried by the pinion E¹, which is rotatably mounted on
the same axis as the numeral wheel. The arrangement of these parts is
such that a rotating motion given any of the pinions E¹, in a clockwise
direction, as shown in the drawings, would give a like action to
their respective numeral wheels; but any motion of the pinions in an
anti-clockwise direction would have no effect on the numeral wheels,
owing to back-stop pawls K, and stop-pins T, provided to allow movement
of the numeral wheels in but one direction.

Co-acting with each pinion E¹, is shown a long lever D, pivoted at
the rear of the machine and provided with a segmental gear rack which
meshes with the teeth of the pinion E¹. This lever comes under what is
now generally termed a segment lever.

Each lever is provided with a spring S, which normally holds the front
or rack end upward in the position shown in Fig. 1, and has co-acting
with it a series of nine depressable keys which protrude through the
casing and contact with the upper edge of the lever.

The arrangement of the keys with their segment levers provides that the
depression of any key will depress the segment lever of that order,
which in turn will rotate the pinion E¹ and its numeral wheel.

While this arrangement is such that each key of a series gives a
different degree of leverage action to the segment lever, and in
turn a degree of rotation to the numeral wheel of the same order in
accordance with the numerical value of the key depressed, it may be
conceived that the momentum set up by the quick stroke of a key would
set the numeral wheel spinning perhaps two or three revolutions, or at
any rate way beyond the point it should stop at to register correctly.

To preserve correct actuation of the mechanism and overcome its
momentum, Felt provided a detent toothed lever for each numeral wheel,
which will be found marked J¹ in the drawings. To this lever he linked
another lever G, which extended below the keys, and arranged the length
of the key-stems so that when each key had revolved the numeral wheel
the proper distance, the key will have engaged the lever G, and through
the link connection will have caused the detent tooth of the lever J¹
to engage one of the pins T, of the numeral wheel, thus bringing the
numeral wheel and the whole train of mechanism to a dead stop.

This combination was timed so that the (1) key would add one, the (2)
key would add two, etc., up to nine for the (9) key. Thus the prime
actuation of each wheel was made safe and positive.

[Sidenote: _Recapitulation of Art prior to Felt calculator_]

Before explaining the means by which the carry of the tens was effected
in the Felt machine without interfering with multiple-order prime
actuation, it will perhaps help the reader to recapitulate on what the
Art already offered.

Going back to the Art, prior to Felt’s invention, there are a few
facts worth reconsidering that point to the broadly new contributions
presented in the Felt invention, and combining these facts with a
little theory may perhaps give a clearer understanding of what was put
into practice.

In most lines of mechanical engineering in the past, the term “theory”
connected with mechanical construction was a bugaboo. But the solution
of the modern calculating machine was wholly dependent upon it.

Let us summarize on the Art, prior to Felt’s invention. A calculating
machine that would calculate, if we eliminate the key-driven feature,
was old. The key-driven feature applied to adding mechanism was old as
adapted to a single-order machine with a capacity for adding only a
single column of digits.

[Sidenote: _Why Hill failed to produce an operative machine_]

Hill attempted to make a multiple order key-driven machine, but failed
because he did not theorize on the necessities involved in the physical
laws of mechanics.

Hill saw only the columnar arrangement of the ordinal division of the
keyboard, and his thought did not pass beyond such relation of the
keys for conveyance. There is no desire to belittle this feature, but
it did not solve the problem that was set forth in the specification
and claims of his patent; neither did it solve it for anyone else who
wished to undertake the making of such a machine.

[Sidenote: _Idiosyncrasies of force and motion increased by use of
keys_]

The introduction of keys as a driving feature in the calculating
machine Art demanded design and construction suitable to control the
new idiosyncrasies of force and motion injected into the Art by their
use, of which the elements of inertia and momentum were the most
troublesome.

[Sidenote: _Light construction a feature_]

Hill, in the design and construction of his machine, ignored these two
elementary features of mechanics and paid the penalty by defeat. The
tremendous speed transmitted to the parts of a key-driven machine,
which has already been illustrated, required that lightness in
construction which is absolutely necessary to reduce inertia to a
minimum, should be observed. The Hill machine design is absolutely
lacking in such thought. The diameter of the numeral wheel and
its heavy construction alone show this. Lightness of construction
also enters into the control of momentum when the mechanism must
suddenly be brought to a dead stop in its lightning-speed action. A
heavily-constructed numeral wheel like that shown in the Hill patent
would be as hard to check as it would to start, even if Hill had
provided means for checking it.

Strength of design and construction, without the usual increase in
weight to attain such end, but above all, the absolute control of
momentum, were features that had to be worked out.

Robjohn partly recognized these features, but he limited the
application of such reasoning to the prime actuation of a single order,
and made nothing operable in a multiple key-driven machine.

Spalding and Bouchet recognized that the application of control was
necessary for both prime actuation and carrying, but, like Robjohn,
they devised nothing that would operate with a series of keys beyond a
single order.

[Sidenote: _Operative features necessary_]

An operative principle for control under prime actuation was perhaps
present in some of the single-order key-driven machines, but whatever
existed was applied to machines with keys arranged in the bank form
of construction, and, to be used with the keys in columnar formation,
required at least a new constructive type of invention. But none of
the means of control for carrying, prior to Felt’s invention, held any
feature that would solve the problem in a multiple-order machine.

[Sidenote: _Classification of the features contained in the early Art
of key-driven machines_]

While all the machines referred to have not been illustrated and
described here, fair samples of the type that have any pertinence
to the Art have been discussed, and those not illustrated would add
nothing more than has been shown. A classification of the inventions
referred to may be made as follows:

Parmelee and Stetner had no carrying mechanism; Hill, Robjohn,
Borland and Hoffman, Swem, Lindholm and Smith had no control for the
carry. Carroll, Bouchet and Spalding show a control for the carrying
action, which in itself would defeat the use of a higher wheel for
prime actuation, and which obviously would also defeat its use in a
multiple-order key-driven machine.

One of the principal reasons why theory was necessary to solve the
problem of the key-driven calculator existed in the impossibility
of seeing what took place in the action of the mechanism under the
lightning speed which it receives in operation. Almost any old device
could be made to operate if moved slow enough to see and study its
action; but the same mechanism that would operate under slow action
would not operate correctly under the lightning-speed action they
could receive from key depression. Only theoretical reasoning could be
used to analyze the cause when key-driven mechanism failed to operate
correctly.

[Sidenote: _Carrying mechanism of Felt’s calculator_]

Referring again to the drawings of the Felt patent, which illustrate
the first embodiment of a multiple-order key-driven calculating
machine, we find, what Felt calls in the claims and specifications,
a carrying mechanism for a multiple-order key-driven calculating
machine. This mechanism was, as set forth in the specification, a
mechanism for transferring the tens, which have been accumulated by
one order, to a higher order, by adding one to the wheel of higher
order for each accumulation of ten by the lower order wheel. This, in
the Felt machine, as in most machines, was effected by the rotation of
a numbered drum, called the numeral wheel, marked with the nine digits
and cipher.

[Sidenote: _Transfer devices_]

The term “transfer device” for such mechanism was in common use, and as
a term it fits certain parts of all classes of devices used for that
purpose, whether for a crank-driven, key-driven, or any other type
of multiple-order or single-order machine. But in the Felt invention
we find it was not the simple device generally used for transferring
the tens. It was, in fact, a combination of devices co-acting with
each other which, in the specification of the patent, was termed the
carrying mechanism.

[Sidenote: _Carrying mechanism versus mere transfer devices_]

Now, carrying mechanism may in a sense be termed a transfer device,
as one of its functions is that of transferring power to carry the
tens, but a mere transfer device may not be truthfully termed a
carrying mechanism for a multiple-order key-driven machine unless it
performs the functions that go to make up a correct carrying of the
tens in that class of machine, and which we find laid down under the
head of carrying mechanism in the Felt patents, where we find the
first operative carrying mechanism ever invented for a multiple-order
key-driven machine.

The functions demanded of such a piece of mechanism are as follows:
First, the storing of power to perform the carry; second, the
unlocking of the numeral wheel to be carried; third, the delivery
of the power stored to perform such carry; fourth, the stopping and
locking of the carried wheel when it has been moved to register such
carry; and fifth, clearing the carrying-lock during prime actuation. A
seemingly simple operation, but let those who have tried to construct
such mechanism judge; they at least have some idea of it and they will
no doubt bow their heads in acknowledgment of the difficulties involved
in this accomplishment.

Mechanism for carrying the tens in single digit adders was one thing,
and such as was used could well be called a transfer device; but
mechanism for carrying the tens in a real key-driven calculating
machine was another thing, and a feature not solved until Felt
solved it, and justly called such combination of devices a “carrying
mechanism.”

[Sidenote: _Details of Felt carrying mechanism_]

In the Felt machine, the carrying mechanism consisted of a lever and
ratchet pawl action, constructed of the parts M, m², operated by a
spring m, the pawl acting upon the numeral wheel pins T, to ratchet
the wheel forward under the spring power. The power in the spring was
developed from the rotation of the lower wheel, which through the means
of an envolute cam[2] attached to left side of each wheel, operated
the carrying lever in the opposite direction to that in which it was
operated by the spring. As the carrying lever passed the highest point
of the cam spiral and dropped off, the stored power in the spring
retracted the lever M, and the pawl m², acting on the higher order
wheel pins T, and moved it one-tenth of a revolution.

[2] NOTE: As all the drawings of the Felt patent are not reproduced
here, the cam is not shown.

This part of the mechanism was in principle an old and commonly-used
device for a one-step ratchet motion used in the carry of the tens.
It served as a means of storing and transferring power from the lower
wheel to actuate the higher wheel in a carrying operation, but a wholly
unqualified action without control.

In the Felt machine a spring-actuated lever N, mounted on the same axis
with the carrying lever, and provided with a detent stop-hook at its
upper end, served to engage the numeral wheel at the end of its carried
action, and normally hold it locked.

An arm or pin P, fixed in and extending from the left side of the
carrying lever and through a hole in the detent lever, acted to
withdraw the detent lever from its locking engagement with the numeral
wheel as the carrying lever reached the extreme point of retraction;
thus the wheel to be carried was unlocked.

Pivoted to the side of the detent lever is a catch O. This catch
or latch is so arranged as to hook on to a cross-rod q, especially
constructed to co-act with the catch and hold the detent lever against
immediate relocking of the numeral wheel as the carrying lever and pawl
act in a carrying motion. The latch has a tail or arm p, which co-acts
with the pin P on the carrying lever in such a way as to release the
latch as the carrying lever finishes its carrying function.

Thus the detent lever N is again free to engage one of the control
or stop-pins T to stop and lock the carried numeral wheel when the
carrying lever and pawl, through the action of the spring stored in the
carrying, has moved the wheel the proper distance.

[Illustration: Bill for First Manufacturing Tools of the “Comptometer”]

A lot of functions to take place in ¹/₁₆₅ of a second, but it worked.
The timing of the stop and locking detents, of course, was one of the
finest features.

[Illustration: Early Comptometer]

The normal engagement of the carrying detent, it may be understood,
would prevent the movement of the wheel by key action or prime
actuation, but the patent shows how Felt overcame this.

The carrying stop and locking detent lever N is provided with a cam-arm
or pin N, which was arranged to co-act with the cam disc E (see Fig.
1), fast to the prime actuating pinion E. The cam surface was short
and performed its function during a short lost motion arranged to take
place before the ratchet pawl would pick up and move the numeral wheel
under key actuation.

The camming action was outward and away from the center, and thus
released the carrying stop from its locking position with the numeral
wheel, and continued rotation of the pinion and cam disc would hold the
lock out of action until the parts had returned to normal.

With the return action of the keys, segment lever, pinion and cam disc,
through the action of a spring attached to the segment lever, the
carrying stop detent will again engage and lock the numeral wheel.

[Sidenote: _Manufacture of the Felt calculator_]

Felt really started to manufacture his calculating machine in the fall
of 1886, after perfecting his invention. Having only a very limited
amount of money with which to produce machines, young Felt, then but
24 years of age, was obliged to make the machines himself, but with
the aid of some dies which he had made for some of the principal parts
(see reproduction of bill for dies on opposite page), he was able
to produce eight finished machines before September, 1887. Two of
these machines were immediately put into service, for the training of
operators, as soon as they were finished.

[Sidenote: _Trade name of Felt calculator_]

Of the first trained operators to operate these machines, which were
given the trademark name “Comptometer,” one was Geo. D. Mackay, and
another was Geo. W. Martin. After three or four months’ practice Mr.
Martin demonstrated one of these machines to such firms as Sprague,
Warner & Co., Pitkin & Brooks, The Chicago Daily News, and the Chicago,
Burlington & Quincy R. R. Co., and finally took employment with the
Equitable Gas Light & Fuel Co. of Chicago (see letter on opposite page)
as operator of the “Comptometer.” The Gas Co. has since been merged
with several other companies into the Peoples Gas Light & Coke Co. of
Chicago.

A very high testimonial of the qualities of the Felt invention was
given by Mr. Martin in 1888, a year after he entered the employment of
the Gas Co., and is reproduced on page 72.

Another fine testimonial was given by Geo. A. Yulle, Secy. & Treas. of
the Chicago Gas Light & Coke Co., in September, 1888 (see page 74). Mr.
Mackay, the other operator, secured employment with Albert Dickinson &
Co., Seed Merchants, as operator of the “Comptometer.” Mr. Mackay was
interviewed a few months ago, and was at that time, after thirty years,
still with the same firm, and a strong advocate of the “Comptometer.”

[Illustration: Letter from Geo. W. Martin]

[Illustration: Testimonial]

[Illustration: Testimonial]

[Illustration]

[Illustration: Letters from Elliott and Rosecrans]

[Sidenote: _Felt calculator Exhibit at National Museum_]

In September, 1887, Felt took one of the first eight machines to
Washington and exhibited it to Gen. W. S. Rosecrans, then Registrar of
the Treasury, and left the machine in the office of Dr. E. B. Elliott,
Actuary of the Treasury, where it was put into constant use. Proof of
the date of this use of Felt’s invention in the Treasury is set forth
in the reproduction of two letters (see opposite page), one was written
by Mr. Elliott and another by Gen. W. S. Rosecrans, in answer to an
inquiry of the Hall Typewriter Co. of Salem, Mass. Another of the first
eight machines was placed with Dr. Daniel Draper, of the N. Y. State
Weather Bureau, New York City.

Felt finally closed a deal with Mr. Robert Tarrant of Chicago, whereby
a partnership contract was signed November 28, 1887. The partnership
was incorporated January 25, 1889, under the name of the Felt & Tarrant
Mfg. Co., who are still manufacturing and selling “Comptometers” under
that name.

[Sidenote: _Significant proof of Felt’s claim of priority_]

Laying aside all the evidence set forth in the foregoing history of
key-driven machines and their idiosyncrasies, significant proof of
Felt’s claim as the first inventor of the modern calculating machine
is justified by the fact that no other multiple-order key-driven
calculating machine was placed on the market prior to 1902.

Lest we lose sight of a most important feature in dealing with the Art
of the Modern Calculator, we should call to mind the fact that as Felt
was the originator of this type of machine, he was also the originator
of the scheme of operation in its performance of the many and varied
short cuts in arithmetical calculation.

The performance of calculation on machines of the older Art differed so
entirely from the new that any scheme of operation that may have been
devised for their use would lend nothing to the derivation of the new
process for operating the key-driven machine of the new Art.

[Sidenote: _Rules for operation an important factor of modern
calculator_]

A superficial examination of one of the instruction books of the
“Comptometer” will convince most any one that it is not only the
mechanism of the machine that made the modern calculator so valuable
to the business world, but also the schemes laid down for its use. The
instructions for figuring Multiplication, Subtraction, Division, Square
Root, Cube Root, Interest, Exchange, Discount, English Currency, etc.,
involved hard study to devise such simple methods and rules.

The instruction books written by Felt for the “Comptometer, the Modern
Calculator,” reflect the genius disclosed in the invention of the
machine itself.

[Illustration]

[Illustration]

[Illustration: From Drawings of Barbour Patent No. 133,188]




Early Efforts in the Recording Machine Art


The Art of recording the addition of columns of figures is old in
principle, but not in practice. Many attempts to make a machine that
would record legibly under all conditions failed. These attempts have
been pointed out from time to time as the first invention of the
recording-adding machine, especially by those desirous of claiming the
laurels.

[Sidenote: _First attempt to record arithmetical computation_]

The first attempt at arithmetical recording for which a patent was
issued, was made by E. D. Barbour in 1872 (see illustration on opposite
page).

E. D. Barbour has also the honor of being the first inventor to apply
Napier’s principle to mechanism intended to automatically register
the result of multiplying a number having several ordinal places by a
single digit without mentally adding together the overlapping figures
resulting from direct multiplication. He patented this machine in 1872
just prior to the issue of his arithmetical recorder patent. (See page
181.)


THE BARBOUR MACHINE

The printing device disclosed in connection with the Barbour machine
for recording calculations was of the most simple nature, allowing only
for the printing of totals and sub-totals.

Its manipulation consisted of placing a piece of paper under a hinged
platen and depressing the platen by hand in the same manner that a time
stamp is used. The ink had to be daubed on the type by a hand operation
to make legible the impressions of the type.

[Sidenote: _Description of Barbour machine_]

The patent drawings of the Barbour machine are so fragmentary that it
is almost impossible to draw any conclusion as to its functions without
reading the specifications.

Fig. 1 represents the base of the machine, while Fig. 4 shows a
carriage which, when in place, is superimposed above the base as
illustrated in Figs. 3 and 5.

The operation of the machine is performed by first pulling out the
slides B (shown in Fig. 1), which set the digital degrees of actuation
of each order; and, second, by operating the hand-lever K, from its
normal position at 0 to 1, if it is desired to add, or to any of
the other numbers in accordance to the value of the multiplier if
multiplication is desired.

The movement of the handle K, from one figure to the other, gives a
reciprocation to the carriage, so that for each figure a reciprocation
will take place.

Each of the slides B, has a series of nine gear racks; each rack has a
number of teeth ranging progressively from 1 tooth for the first gear
rack to 9 teeth for the last rack, thus the pulling out of the slides B
will present one of the gear racks in line to act upon the accumulator
mechanism of the carriage as the carriage is moved back and forth over
it.

The accumulator mechanism consists of the register wheels M¹ and M²
and the type wheels M³ and M⁴ mounted on a common arbor and a carry
transfer device between the wheels of each order.

Operating between the accumulator wheels and the racks of plate B are
a pair of gears, one in the form of a lantern wheel loosely mounted on
the accumulator wheel shaft but connected thereto by a ratchet wheel
and pawl connection; the other, a small pinion meshing with the lantern
wheel on a separate axis, protrudes below the carriage into the path of
the racks.

Thus as the carriage is moved by the reciprocating device connected
with the hand-lever K, the pinions of the accumulator will engage
whatever racks have been set and the numeral wheels and type wheels
will be operated to give the result.

The numeral and type wheels have two sets of figures, one of which is
used for addition and multiplication, while the other set runs in the
opposite direction for negative computation or subtraction and division.

A plate arranged with sight apertures covers the numeral or register
wheels, while the type wheels are left uncovered to allow a hinged
platen F, mounted on the top of the carriage (see Fig. 3), to be swung
over on top of them and depressed.

Attached to the platen F, are a series of spring clips d, under which
strips of paper may be slipped (as shown by D, in Fig. 4), and which
serves to hold the paper while an impression is taken.

[Sidenote: _Barbour machine not practical_]

Thus the Barbour invention stands in the Art as something to show that
as early as 1872 an effort was made to provide means to preserve a
record of calculations by printing the totals of such calculations.


THE BALDWIN MACHINE

The next effort in this class of machines is illustrated in a patent
issued to Frank S. Baldwin in 1875 (see illustration on opposite page).
The Baldwin machine is also of moment as having the scheme found in
the machines known as the Brunsviga, made under the Odhner patents--a
foreign invention, later than that of Baldwin, used extensively abroad
and to a limited extent in this country.

The contribution of Baldwin to the Art of recording-calculating devices
seems to be only the roll-paper in ribbon form and the application
of the ink ribbon. The method used by Barbour for type impression
was adapted and used by Baldwin; that is, the hinged platen and its
operation by hand.

Of the illustrations shown of the Baldwin machine, one is reproduced
from the drawings of the patent while the other is a photo reproduction
of the actual machine which was placed on the market, but, as may be
noted, minus the printing or recording device shown in the patent
drawings.

[Sidenote: _Description of Baldwin machine_]

Referring to the photo reproduction, the upper row of figures showing
through the sight apertures in the casing are those of the numeral
wheels which accumulate the totals, and which in the patent drawings
would represent the type of the accumulator wheels for printing the
totals of addition and multiplication or the remainders of subtraction
and division.

[Illustration: From Drawings of Baldwin Patent No. 159,244]

[Illustration: Baldwin Machine]

The figures showing below serve to register multiples of addition and
subtraction which would read as the multiplier in multiplications
or the quotient in division. These wheels are the type wheels N, in
the patent drawings, which serve the purpose of recording the named
functions of calculation.

The means by which the type wheels of the upper row are turned through
the varying degrees of rotation they receive to register the results
of calculation, consists of a crank-driven, revolvable drum, marked E,
which is provided with several denominational series of projectable
gear teeth h, which may be made to protrude through the drum by
operation of the digital setting-knobs g, situated on the outside of
the drum.

These knobs, as shown in the patent drawings, are fast to radial arms,
each of which serves as one of three spokes of a half-wheel device,
operating inside the drum and pivoted on the inner hub of the drum.

These half wheels marked F, in the drawings, by means of their cam
faces h¹, serve to force the gear teeth out through the face of the
drum, or let them recede under the action of their springs as the knobs
g, are operated forward and back in the slots x, of the drum provided
for the purpose.

As will be noted from the photographic reproduction of the machine,
these slots are notched to allow the arms extending through them to
be locked in nine different radial positions, and that each of these
positions are marked progressively from 0 to 9.

This arrangement allows the operator to set up numbers in the different
orders by springing the setting-knobs g to the left and pulling them
forward to the number desired, where it will become locked in the notch
when released. This action will have forced out as many gear teeth
in each order as have been set up by the knobs g in their respective
orders.

The lateral positions of the projectable gear teeth correspond to
the spacing of the type-wheels, and an intermediate gear G, meshing
with each type, or register wheel, is loosely mounted on the shaft H,
interposed between the said wheels and the actuating drum E, so that
when the drum is revolved by the crank provided for that purpose, the
gear teeth protruding from the drum will engage the intermediate gears
G, and turn them and their type or register wheels as many of their ten
points of rotation as have been set up in their respective orders of
the setting devices of the drum.

Revolving the drum in one direction adds, while revolving it in the
opposite direction subtracts, and repeated revolutions in either
direction give respectively the multiple forms of addition or
subtraction which result in either multiplication or division, as the
case may be.

The actuating drum E, is provided with means by which it may be shifted
to the left to furnish means for multiplying by more than one factor
and to simplify the process of division.

The means for the carry of the tens consist of a series of teeth i,
formed by the bent end of a pivoted spring-pressed lever arm which is
pivoted to the inside of the actuating drum with the tooth protruding
through a slot in the drum, so arranged as to allow motion of the tooth
in a direction parallel to the drum axis.

[Illustration]

[Illustration: From Drawings of Pottin Patent No. 312,014]

Normally these teeth are held in a position to escape engagement with
the intermediate gears G, but provision is made for camming the teeth
i, to the left into the path of an intermediate gear of one order as
the type or register wheel of the lower order passes from 9 to 0.

The parts which perform this function are the cam m, located on the
left side of each wheel, the plunger M, which operates in the fixed
shaft H, and which has a T-shaped head that, when projected into the
path of the carrying teeth i, serve to cam them sidewise and bring
about the engagement referred to, which results in the higher type or
numeral wheel being stepped forward one space.

The cam-lugs j on the drum serve to engage and push back the T heads of
the cam plungers M, after they have brought about the one-step movement
of the higher wheel.

[Sidenote: _Baldwin’s printing mechanism_]

The printing device consists of a hand-manipulated frame pivoted to
the main frame of the machine by the shaft t. The paper is supplied
from a roll about the shaft t, and an ink-ribbon is fed back and forth
from the rolls u and u¹ over bars of the printing-frame which protrude
through slots in the casing and act as platens for the impression of
the paper and ink-ribbon against the type.

It is presumed that the paper was torn off after a record was printed
in the same manner as in the more modern machines.


THE POTTIN MACHINE

Eight years after the Baldwin patent was issued, a Frenchman named
Henry Pottin, residing in Paris, France, invented a machine for
recording cash transactions, which he patented in England in 1883 and
in the United States in 1885 (see illustration on opposite page).

The form and design of the machine, as will be noted, correspond quite
favorably with the scheme of the present-day cash register, although it
lacks the later refinement that has made the cash register acceptable
from a visible point of view.

[Sidenote: _First key-set crank-operated machine and first attempt to
record the items in addition_]

The Pottin invention is named here as the first in which two of the
prime principles of the recording-adders of today are disclosed; one
is the depressable key-set feature and the other is the recording of
the numerical items. The Pottin machine was the first known depressable
key-set crank-operated machine made to add columns of figures and the
first machine in which an attempt was made to print the numerical items
as they were added.

Turning to the illustration of the U. S. patent drawings of the Pottin
machine, the reader will note that there are four large wheels shown,
marked B. These wheels are what may be called the type-wheels, although
they also serve as indicator wheels for registering cash sales. The
type figures are formed by a series of needles fixed in the face of the
wheels.

The means employed for presenting the proper type figure for printing
and likewise the indicator figures to indicate the amount set up in
each denominational order was as follows:

Referring to Fig. 1, it will be noted that to each type-wheel is geared
a spring-actuated segmental rack marked D, which, as shown in the
drawing, is in contact with a pin marked i, which protrudes from the
side of the depressed number (9) key.

The normal position of the rack D, is indicated in dotted lines showing
the next higher sector which has not been displaced by key depression.

[Sidenote: _Description of Pottin machine_]

Each key, as will be noted from Fig. 7, is provided with one of the
pins i, which is normally out of the path of the lug j, as the racks D,
drop forward; but when any key is depressed the pin is presented in the
path of the lug j, and stops further forward action of the rack.

It will be noted that the arrangement of the keys is such as will allow
progressively varying degrees of action to the segmental racks D. This
variation, combined with the geared relation of the type-wheels and
racks is equivalent to a tenth of a rotation of the type-wheel for each
successive key in the order of their arrangement from 1 to 9.

The means provided for holding the segmental racks D, at normal, also
serves to hold a key of the same order depressed, and consists of a
pivoted spring-pressed latch-frame marked E (see Figs. 7 and 8).

With such a combination, the depression of keys in the several orders
will unlatch the segmental racks, and the racks, through the tension
of their actuating springs, will turn the wheels and present a type
corresponding to the numerical value of each key depressed.

A hand lever, marked R, located on left side of the machine provides
power for printing the items. Another hand lever, marked J, serves
to restore the segmental racks, type-wheels and the keys to normal,
and through the co-operation of the lever R, adds the items to the
totalizer numeral wheels, which are shown in Fig. 1 as the numbered
wheels marked v.

The paper is supplied from a roll mounted on a hinged platen frame P¹,
supported in its normal position by a spring P³. The paper passes under
the roller P, which acts as a platen for the impression of the type. A
shaft Q, passing under the frame P¹, is fast and rigidly connected on
the left-hand side of the machine with the hand lever R, and acts as a
pivot for the said lever and by means of lateral projections q, serves
when the lever R is operated to engage the frame P¹, and depresses it
until the needle types have pricked the numerical items through the
paper.

A slit in the casing provided means for printing the item on a separate
piece of paper or bill.

Although there is no means shown by which the paper is fed after an
item is printed, it is claimed in the specification that the well-known
means for such feeding may be employed. The actuating lever J referred
to, is connected by a ratchet and geared action with the shaft F[3],
so that a revolution is given the said shaft each time the lever is
operated.

[3] NOTE: All the drawings of the Pottin patent are not shown here.

To the shaft F, (see Fig. 1) is attached a series of arms H, one for
each order, which, as the shaft revolves in the direction of the arrow,
engages a lug marked I, on the segmental racks D, thus rocking the
segments back to normal, turning the type-wheels with them.

The return of the segment racks D, cause the back of the latch-tooth
f¹, (see Fig. 8) to engage the latch-tooth f, of the latch bar E,
camming it out of engagement with the keys so that any key that has
been set will return by means of its own spring.

[Illustration]

[Illustration: From Drawings of Burroughs Patent No. 388,118]

[Illustration: Wm. S. Burroughs]

The total or accumulator numeral wheels are connectable with the type
or indicating wheels B, by an engaging and disengaging gear motion set
up by the combined action of the hand levers R and J, which first cause
such gear engagement, and then, through the return of the type wheels
to zero, turn the accumulator wheels, thus transferring the amount of
the item set upon the type wheels to the accumulator wheels.

The specification claims the machine is intended for use by cashiers,
bank-tellers, and others, to record receipts or disbursements.

It is also claimed in the specification that instead of the needle type
ordinary type may be used in combination with an inking ribbon if so
desired.

[Sidenote: _Early efforts of Wm. S. Burroughs_]

One of the next attempts to produce a recording-adder was made by Wm.
S. Burroughs, whose name sixteen years later was used to rename the
American Arithmometer Co., now known as the Burroughs Adding Machine Co.

The first patent issued to Burroughs, No. 388116, under date of August
21, 1888, like the machine of Barbour and Baldwin, was designed to
record only the final result of calculation.

On the same date, but of later application, another patent, No. 388118,
was issued to Burroughs which claimed to combine the recording of the
numerical items and the recording of the totals in one machine. Some of
the drawings of this patent have been reproduced. (See opposite page.)


MACHINE OF EARLY BURROUGHS PATENT

Referring to the drawings of the Burroughs patent, it will be noted,
that in outward form, the machine is similar to the Burroughs machine
of today. To give a detailed description of the construction of the
machine of this Burroughs patent would make tedious reading and take
unnecessary space.

[Sidenote: _General scheme of Burroughs’ first inventions_]

The principle involved in the mechanism for recording the items is very
similar to that of the Pottin invention; the setting of the type wheels
being effected as in the Pottin machine by means of segment gears which
the depression of the keys serves to unlatch, and acts to gauge the
additive degree of their movement.

Burroughs used the inking form of type proposed as an alternative by
Pottin in his patent specification instead of the needles shown in the
Pottin drawings.

In the Burroughs patent, as in the Pottin, it will be noted that there
are two sets of wheels bearing figures, one set of which, marked J,
situated at the rear, are the type-wheels, and the other set, marked A,
at the front of the machine, are for the accumulation of the totals.

For each denominational order of the type and total wheels, there
is provided an actuating segmental gear, consisting of a two-armed
segmental lever pivoted to the shaft C, and having the gear teeth of
its rear arm constantly in mesh with the pinion gear of the type-wheel
J, and the gear teeth of the forward arm normally presented to, but out
of mesh with the pinion gear of its total wheel A.

Each of these denominational actuators or segment gears is provided
with a stop projection X², at the top end of its forward gear rack,
which serves as a means for interrupting the downward movement of
that end of the segment lever, and thus controls its movement as a
denominational actuator.

It will be noted that instead of the key-stems acting directly as a
stop for the denominational actuators, as in the Pottin invention,
Burroughs used a bell crank type of key lever and the stop-wire C¹ as
an intermediate means, and in this manner produced a flat keyboard more
practical for key manipulation.

[Sidenote: _Brief description of machine of early Burroughs patents_]

The stop-wires C¹, as will be noted, are arranged to slide in slots
of the framework, and while normally not presented in the path of the
stop-projection X², of the denominational actuators, it may be observed
that by the depression of the proper key any one of them may be drawn
rearward and into the path of the stop projection X², of its related
actuator, and thus serve as a means to intercept the downward action of
the actuator.

The denominational actuators in the Burroughs machine were not provided
with spring tension that would cause them to act as soon as unlatched
by depression of the keys as has been described in relation to the
Pottin invention.

While the keys in the Burroughs machine, as in the Pottin invention,
served also to unlatch the denominational actuators in their respective
orders, no movement of the said actuators or type-wheels took place
until a secondary action was performed.

The secondary action, or the operation of the hand lever, marked C⁵,
attached to the shaft C, on its initial or forward stroke dragged the
denominational actuators down by means of friction and thus set the
type-wheels, and by means claimed in the specification, brought about
the type impression to print the result of the key-setting or the item
so set.

The backward or rear stroke of the hand lever caused the accumulator or
total numeral wheels to be engaged and the item to be added to them.

From this single lever action it will be noted that there is an
improvement shown over and above the Pottin invention in the fact that
but one lever motion is required; Pottin having provided two levers so
that in the event of error the operation of one lever would reset the
machine without performing any addition or printing.

In the Burroughs invention, the motion of denominational actuators and
their type-wheels not being effected through depression of keys, as in
the Pottin machine, allowed any error in the setting up of an item to
be corrected by the resetting of the keys and relatching of the gears,
which it is claimed was provided for by operation of the lever marked
B⁷ (Fig. 1 of the drawings).

As a means of supplying power to his denominational actuators,
Burroughs provided what may be called a universal actuator common to
all orders, composed of a rock-frame (arms D², loose on each end of
actuating shaft C, and having their outward ends rigidly connected by
the bar a⁹) and the arms E, fixed to each end of the shaft C.

Projecting from the inside of each of the arms E, are two lugs, b¹ and
b³, which contact with the arms D² of the rock-frame as the shaft C is
rocked back and forth by its hand crank C⁵, and thus lower and raise
the rock-frame.

The means employed to transmit the reciprocating action of the
universal actuator to such denominational actuators as may be unlatched
by key depression, consists of a series of spring-pressed arc-shaped
levers D¹, pivoted to the rock-frame bar a⁹, which bear against a pin
b² fixed in the front arm of the denominational actuators.

Each of the levers D¹, is provided with a notch y, which serves on
the downward action of the rock-frame to engage the pins b², of the
denominational actuators and draw down with them such actuators as have
been unlatched by key depression and to pass over the pins of such
actuators as have not been unlatched.

When in the course of such downward movement the denominational
actuators are intercepted by the stop-wires C¹, the yielding spring
pressure of the levers D¹, allow the notches y, to slip over the pins
b², and leave the denominational actuators and their type-wheels set
for recording the item thus set up.

The means provided for impression of the type is shown in other
drawings of a patent not reproduced here. The means provided consisted
of a universal platen, which, the specification states, serves to press
the ink-ribbon and paper against the type after all the figures of each
item were set.

While Barbour, Baldwin and Pottin all used the universal platen to
print the collective setting of type represented in the items or
totals, as the case may be, each varied somewhat in detail. Baldwin
used a toggle to press the platen toward the type, while Burroughs used
a spring to press the platen against the type and a toggle to press it
away from the type.

Burroughs claimed to have combined in his invention the printing of
the totals, with the printing of the items, each of which it has been
shown was claimed by the patentees of previous inventions but had not
been combined in one machine prior to the Burroughs attempt.

The process for recording these totals in the Burroughs patent
consisted of utilizing the action of the total wheels during their
resetting or zeroizing movement to gauge the setting of the type-wheels.

The specification shows that, during the downward motion or setting
of the denominational actuators, as they set the type wheels, the
numeral wheels are out of gear and receive no motion therefrom; and
that after the recording of each item and during the return motion of
denominational actuators, the numeral or total wheels are revolved
forward in their accumulative action of adding the items and thus
registering the total.

Provision is made, however, when it is desired to print the totals, to
cause the totalizing wheels to enmesh with the denominational actuators
on their downward or setting movement, and for the unlatching of all
the racks so that by operating the hand lever C⁵, the downward action
of the racks will reverse the action of the totalizing wheels, which
will revolve backward until the zeros show at the visible reading
point, where they will be arrested by stops provided for that purpose.
By this method the forward rotation accumulated on each wheel will,
through the reverse action of zeroizing, give a like degree of action
to the type-wheels through the denominational actuators. Thus the
registration of the total wheels, it is claimed, will be transferred
to the type-wheels and the record printed thereof as a footing to the
column of numerical items that have been added.

[Sidenote: _All early arithmetical printing devices impractical_]

To pass judgment on the recording machines of the patents that have
been described, from the invention of Barbour to that of Burroughs,
demands consideration, first, as to whether in any of the machines of
these patents the primary features of legible recording were present.

The question as to operativeness respecting other features is of no
consideration until it is proven that the means disclosed for recording
was practical. As non-recording adding or calculating machines they
were not of a type that could compete with the more speedy key-driven
machines dealt with in the preceding chapters; therefore without
the capacity for legible recording, these patents must stand as
representing a nonentity or as statutory evidence of the ineffective
efforts of those who conceived the scheme of their make-up and
attempted to produce a recording-adding machine.

Without the capacity for legible recording, of what avail is it that
the machine of one of these patents should disclose advantages over
another? It may be conceded that there are features set forth in the
Pottin and Burroughs patents that if operatively combined with legible
recording would disclose quite an advanced state of the Art at the time
they were patented. But credit for such an operative combination cannot
be given until it exists.

There is no desire to question the ingenuity displayed by any of
these inventors, but in seeking the first practical recording-adding
or calculating machine we must first find an operative machine of
that type; one which will record in a practical and legible manner
regardless of its other qualifications.

[Sidenote: _Practical method for recording disclosed later_]

The fact that the fundamental principle used for the impression of the
type in the practical recorder of today is not displayed in any of
these inventions, raises the question as to the effective operativeness
of the printing scheme disclosed in the patents of these early machines.

In each of the four alleged recording-adding machine patents described,
it will be noted that the means employed for printing was that of
pressing the paper against the group of type by means of a universal
platen or plate.

While with such a combination it may be possible to provide a set
pressure great enough to legibly print a numerical item or total having
eight to ten figures through an ink ribbon, it would not be practical
to use the same pressure to print a single-digit figure, as it would
cause the type to break through the paper. And yet in the numerical
items and totals that have to be recorded in machines of the class
under consideration, such wide variation is constantly encountered.

We are all familiar with the typewriter and the legible printing it
produces. But suppose instead of printing each letter separately the
whole word should be printed at once by a single-key depression,
then, of course, single-letter words, such as the article “a” or the
pronoun “I” would also have to be printed by a single-key depression.
In this supposition we find a parallel of the requirements of a
recording-adding machine.

[Illustration]

[Illustration: Drawings of Ludlum Patent No. 384,373]

[Sidenote: _Inoperative features of early recording mechanism_]

If it were possible to so increase the leverage of the typewriter
keys enough to cause a word of ten letters to be printed as legibly
as a single letter is now printed, ten times the power would have
to be delivered at the type-head. Then think what would happen with
that same amount of power applied to print the letter “a,” or letter
“I.” You would not question that under such conditions the type would
break a hole in the paper. And yet the patentees of the said described
inventions wanted the public to believe that their inventions were
operative. But to be operative as recording-adding machines, they must
meet such variable conditions as described.

It is useless to believe that a variation of from one to ten or more
type could be printed by a set amount of pressure through an ink-ribbon
and be legible under all circumstances.

While the needle-type of Pottin may have printed the items legibly
enough for a cash register, it would not serve the purpose of a record
for universal use. The use of regular type and the inking ribbon
proposed in his specification would bring it within the inoperative
features named.


THE LUDLUM MACHINE

In 1888, about two months prior to the issue of the Burroughs recording
machine patent just referred to, a patent was issued to A. C. Ludlum
for an adding and writing-machine. (See illustration on opposite page.)

[Sidenote: _Adding mechanism attached to typewriter_]

It will be noted by reference to the drawings that the scheme is that
of a typewriter with an adding mechanism attached.

The details of the typewriter may be omitted, as most of us are
familiar with typewriters. A feature that differed from the regular
typewriter, however, was that the machine printed figures only and the
carriage operated in the opposite direction, thus printing from right
to left instead of left to right.

[Sidenote: _Description of Ludlum machine_]

A series of numeral wheels and their devices for the transfer of the
tens, designed to register the totals, are shown mounted in a shiftable
frame connected with the bar marked F, with the typewriter carriage,
and is claimed to move therewith.

Each numeral wheel is provided with a gear marked G, which, as the
carriage moves after writing or printing each figure of the item, is
supposed to slide into mesh one at a time with an adding gear marked
H, the engagement taking place from right to left. Or beginning with
the right or units numeral wheel a higher order numeral wheel gear is
supposed to shift through movement of the carriage into engagement with
the adding gear H, each time a key is depressed.

The adding gear H, is supposed to receive varying degrees of rotation
from the keys according to their numerical marking and to rotate the
numeral wheel with which it happens to be engaged, a corresponding
number of its ten marked points of registration.

Between the adding gear H, and the keys which act to drive it, is a
ratchet and gear device consisting of the ratchet pawl pivoted to the
adding gear H, the ratchet I⁶, and its pinion gear, the segment gear
I² fast to the rock shaft I, the nine arms I¹ fast to the rock shaft
and the pins I², which are arranged in the key levers to contact with
and depress the arms I¹ of the rock shaft varying distances, according
to the value of the key depressed. That is, supposing that the full
throw of the key-lever was required to actuate the rock shaft with
its gear and ratchet connection to give nine-tenths of a revolution
to the numeral wheel in adding the digit nine, the pin I² in the (9)
key-lever would in that case be in contact with its arm I¹, of the
rock shaft, but the pins I², of each of the other key levers would be
arranged to allow lost motion before the pin should engage its arm I¹
of the rock shaft, in accordance with the difference of their adding
value.

According to the specification, Ludlum evidently had the idea that he
could stop the adding gear H, while under the high rate of speed it
would receive from a quick depression of a key, by jabbing the detent
J between the fine spacing of the gear teeth shown in his drawing.
But to those familiar with the possibility of such stop devices, its
inoperativeness will be obvious; not that the principle properly
applied would not work, for its application by Felt prior to that of
Ludlum proved the possibilities of this method of gauging additive
actuation.

The detent lever J, as shown in the drawings, is operated by the hinged
plate D, through action of the key levers, as any one of them are
depressed.

Under depression of a key, the hinged plate D, being carried down with
it, engages the arm J³ of the detent and throws the tooth at its upper
end into the teeth of the gear H.

The timing of the entry of the tooth of the detent is supposed to be
gauged to enter the right tooth, but as the action of these parts is
fast, slow or medium at the will of the operator, considerable time
must be allowed for variation in the entry of the detent tooth, which
requires space, as certain parts will fly ahead under the sudden impact
they may receive from a quick stroke, where they would not under a slow
stroke, but no allowance was provided for such contingency.

The means provided for the carry of the tens consist of the gears G⁹,
meshing with the numeral wheel gears and the single gear tooth g⁹,
attached to it, which, at each revolution of the lower wheel, as it
passes from 9 to 0, engages the gear of the numeral wheel of higher
denomination and was supposed to turn the higher gear one-tenth of a
revolution, thus registering one greater.

On account of the Gears G⁹, of one order and the gear tooth g⁹, of
another order operating on the same numeral wheel gear, the transfer
gears are arranged alternately on separate shafts, one at the side and
one below the numeral wheels.

[Sidenote: _Ludlum machine inoperative_]

The mechanical scheme disclosed in the Ludlum patent, to the
unsophisticated may seem to be operative. But to those familiar with
the Art of key-driven adding mechanism it will at once be obvious that
even if the typewriter feature was constructed properly the possibility
of correctly adding the items as they were printed was absolutely
impossible.

Laying aside several other features of inoperativeness, obvious to
those who know such mechanism, the reader, although not versed in the
Art of key-driven adding mechanism, will observe from the preceding
chapter, that the means provided for transferring the tens without
any control for the numeral wheels against over-rotation, would make
correct addition impossible.

The drawings and specification of the Ludlum patent disclose a mere
dream and show that they were not copied from the make-up of an
operative machine.

It was a daring scheme and one that none but a dreamer would undertake
to construct in the method shown. There have in later years been some
successful ten-key recording machines made and sold, but they were of a
very different design and principle.

There have also been several adding attachments made and sold that
could be adjusted to a regular commercial typewriter that are claimed
to be dependable, but none of these machines were early enough to be
claimed as the first operative recording-adding machine, or the first
adding machine in which the principle used for the legible recording of
the numerical items used in the machines of today may be found.

[Illustration]




FIRST PRACTICAL RECORDERS


The fact that Barbour, Baldwin, Pottin, Ludlum and Burroughs attempted
to produce a recording-adding machine shows that as far back as 1872,
and at periods down to 1888, there was at least in the minds of these
men a conception of the usefulness of such a machine, and the fact that
there were five with the same thought is fairly good evidence of the
need for a machine of this class.

[Sidenote: _Burroughs a bank clerk_]

In some of the human-interest articles issued through the advertising
department of the Burroughs Adding Machine Co. it is stated that
Wm. Seward Burroughs was a bank clerk prior to his efforts at
adding machine construction. It is conceivable, therefore, that his
first efforts at adding machine invention should be directed toward the
production of a machine that would be of service in the bank for the
bringing together of the loose items of account that are to be found in
the form of checks, drafts, and the like, by printing a record of the
items and their totals during the process of adding them together.

[Sidenote: _Felt interested in recorder Art_]

It is not surprising, therefore, that a manufacturer of a successful
calculating machine should, through his contact with the trade, come
to the conclusion that there was use for a machine of this class
in the banks. As proof of this, we find that an application for a
recording-adding machine patent was filed January 19, 1888, by D. E.
Felt, which was allowed and issued June 11, 1889.

[Illustration]

[Illustration: From Drawings of Felt Patent No. 405,024]

[Sidenote: _Felt’s first recording machine_]

Some of the drawings of this patent will be found reproduced on the
opposite page, from which the reader will note that Felt combined his
scheme for recording with the mechanism of the machine he was then
manufacturing and selling under the trade name of “Comptometer.”

In this patent is shown the first application of the type sector
combined with the individual type impression for printing the figures
of the items as they were added, thus giving equal impression, whether
there were one or a dozen figures in the item or total to be printed.

While the average mechanical engineer would not at a glance recognize
any great advantage in placing the type figures directly on the sector
instead of using the type-wheel and segment gear to drive it, as shown
in two of the previously described patents, there is plenty of evidence
of its advantage in the fact that all the later successful inventors
have followed the Felt scheme. It provided more simple construction for
the narrow space these parts must occupy for practical linear spacing.

[Sidenote: _Fell recording mechanism combined with his calculating
machine_]

As the adding mechanism of this machine corresponds to that of the Felt
patent 371,496, previously described in the preceding chapter, it is
not necessary to duplicate the description here. Suffice it to say,
that by the depression of a key in any order, the value of that key is
added to the numeral wheel of that order, and if the figure added is
great enough when added to that previously registered on the wheel, a
ten will be transferred to the higher wheel by a carrying mechanism
specially provided to allow the said higher wheel being in turn
operated by an ordinal series of keys, thus providing the means whereby
a series of denominational orders of key-driven adding mechanism may be
interoperative.

[Sidenote: _Description of Felt’s first recorder_]

In Fig. 2 of the drawings is shown the result of striking the (8) key,
which may be considered illustrative of such action in any order,
whether units, tens, hundreds, thousands, etc.

The depression of the (8) key is shown to have carried the lever D down
eight of its nine additive points of movement, causing the plunger 15,
bearing against its upper edge, to drop with it under the action of the
plunger spring 17.

To the upper end of this plunger, is pivotally attached an arm of
the type sector U, which is in turn pivoted to the rod y, and by
the lowering of the plunger 15, is rocked on its pivot, raising the
type-head until the number (8) type is presented opposite the printing
bar or platen T, which is hung on the pivot arms T¹, so that it may be
swung forward and backward.

An ink-ribbon w, and its shifting mechanism is provided, as shown in
Fig. 1; the paper v, is supplied in ribbon form from a roll and passes
between the ink-ribbon and the platen T.

Normally, the platen, the paper and the ink-ribbon are in a retracted
position, allowing space for the type sector to raise and lower freely.
But, as shown in Fig. 2, a type impression is taking place through the
escapement of the cam wheel R¹ which is located back of the platen,
and which, as shown, has forced the cam lever 1 forward, pressing the
spring p, against the platen T, thus forcing the paper and ribbon
forward against the type, and printing the figure 8.

After the cam-tooth passes, the platen, paper, ink-ribbon and spring
return to normal, allowing the type sector freedom to drop when the key
is released.

The cam wheel R is propelled by a spring S (Fig. 1), wound by the
hand-knob S³, and is released for action through the escapement of the
pallet wheel R attached to the cam wheel R and the pallet c.

The pallet c is tripped each time a key is depressed and is shown in
the tripped position operated by the link P and the plural-armed lever
O, N, which through its manifold arms N, may receive action through
pins a, of any of the rock bars L, as they are depressed by the keys.

The cycle of action described takes place with every key depressed,
except that the movement of the type sector varies according to the key
depressed.

[Sidenote: _First individualized type impression combined with printing
sector_]

As the printing in this Felt invention was by individualized type
impression, legibility of recording as well as accurate addition was
obtained. Although this patent shows that Felt had produced such an
operative combination, there are two features in this patent which
would prevent its becoming a marketable machine.

One of these features was that of having to wind the motor spring that
furnished power for the type impression. The other feature was that
there was no provision for printing the ciphers. Although the ciphers
were always omitted from the keyboard of non-recording adders, as they
could perform no function in addition or other forms of calculation,
they could not without inconvenience, be eliminated from items in
recording.


THE SECOND FELT RECORDER

[Sidenote: _First practical arithmetical recorder_]

While the last-described Felt patent was still pending, Felt improved
his mechanism for recording, installing new features and eliminating
the objectionable features referred to. These improvements were of
such a satisfactory nature that the Felt & Tarrant Mfg. Co. made
twenty-five recording-adders, with the new features, which were sold to
various banks. The first of these machines was placed on trial with the
Merchants & Manufacturers National Bank of Pittsburgh, Pa., in December
of 1889.

Good evidence of the practical features of this machine was set forth
in a testimonial given at the time by W. A. Shaw, the cashier of the
bank, after it had been given a six months’ test. This testimonial is
extant and has been reproduced on opposite page.

[Sidenote: _The first sale of a recording-adding machine on record_]

Records show that the bank purchased that “Comptograph,” which was the
trade name given the Felt recording-adder, and used it until 1899, at
which time this machine, along with others of the same make purchased
at a later date, were replaced by the bank with “Comptographs” of more
modern type.

This Felt recording machine was without question the first practical
recording-adding machine ever sold that would produce legible printed
records of items and totals under the variable conditions that have to
be met in such a class of recording.

[Illustration: Testimonial]

[Illustration: Felt Recording and Listing Machine.

Purchased and Used for Ten Years by the Merchants & Manufacturers Bank
of Pittsburgh, Pa.

Machine is now in the National Museum at Washington]

After ten years of service this first practical recording-adding
machine was still in excellent condition, and in 1907 was secured
by the Comptograph Co. from the Bank of Pittsburgh, into which the
Merchants & Manufacturers National Bank, along with other banks, had
been merged. It was finally procured by Mr. Felt and presented to
the National Museum of Washington, D. C., where it may now be found
on exhibit along with other inventions produced by Felt. A photo
reproduction of this machine as it appeared before it was presented to
the Museum, is shown on the opposite page.

[Sidenote: _Features of first practical recorder_]

Like the machine of the first Felt recorder patent, it was a visible
printer, each figure being printed as the key was depressed, the paper
being shifted by the hand lever shown at the right.

Unlike the former machine, however, the operator was not called upon to
perform the extra operation of winding up a spring to furnish power for
the printing.

Power for the printing was stored by the action of the paper
shift-lever and an entirely different printing device was used.
Provision for printing the ciphers automatically was also a feature of
this machine. It was not necessary to operate cipher keys, and there
were no such keys to be operated. To print an item having ciphers in
it required only the omission of the ciphers as the ciphers would
automatically fill in.

The arrangement of the paper shows a good improvement over the first
machine, as it was more accessible, being fed from a roll at the top
down and around rolls below and looped back so that it is moved upward
on the printed surface, where it may be torn off as desired.

The mechanism of this machine is not illustrated in any one patent.
The Felt patents Nos. 441,233 and 465,255 cover the new feature, but
the later patent, No. 465,255, shows it best. Some of the drawings of
the last-named patent are reproduced on the opposite page to help in
explanation of the details of the new features.

[Sidenote: _Description of Felt’s second recorder_]

By referring to the drawings, it will be noted that the form of the
front of the casing differs from the machine. Other drawings of the
patent, not shown here, disclose features of still later invention
than were in the machine of the photo reproduction. But it is with the
printing device that we are now interested, and it was in this patent
that it was first shown in the form used in the first marketed machine
referred to.

The type sector marked 81 is like that of the first patent, except that
it is provided with the ciphers as well as the nine digits.

The cipher type are always presented for printing when the sectors
are resting at normal. Thus, if an impression can be made without
depressing the keys in that order, a cipher will be printed, as will be
shown later.

Back of the paper and pivoted to the rod 97, are a series of printing
hammers 87, one for each type sector.

The hammers are operated by the spring 88, and are shown retained
against the tension of their springs by the trigger latches 89.

These trigger latches are pivoted on the fixed shaft 171ᵃ, and actuated
by the springs 92 to cause their engagement with the notch 90 of the
printing hammers.

[Illustration: From Drawings of Felt Patent No. 465,255]

Each of the trigger latches are provided with a laterally extending lug
93, formed on their lower arm, and each lug overlaps the back of the
lower arm of the adjacent trigger latch to the right of it, so that if
any trigger latch should be operated so as to extricate it from the
notch 50 of its printing hammer, its overlapping lug 93, would cause a
like action of the trigger latch to the right of that, and so on; thus
releasing all the trigger latches to the right of the latch originally
released. Such releasing, of course, allowed the printing-hammers 87,
to spring forward in all the orders so affected.

The long-stop actuating lever marked 16, corresponds with the lever
G of the Felt key-driven calculator shown in a preceding chapter,
and performs the same function as the rock bars L of the first Felt
recorder patent. These stop levers 16 are pivoted at 17, and are
provided with rear arms 86, extending upward with their ends opposite
the lateral extending lug 93, of the trigger latch, which corresponds
to the order of keys which the lever 16 serves.

In the rear upwardly-extending end of each of these levers 16, an
adjusting screw 91, is provided as a tappet for tripping the trigger
latch corresponding to its order.

From the above-described combination of mechanism, it may be seen that
if a key in any order is depressed, it will, as it comes in contact
with the stop lever 16, not only cause the adding mechanism to be
stopped through the stop 19, but it will also, through its rear arm
86, cause the trigger latch of its order to trip, and likewise all the
trigger latches and printing-hammers to the right, thus printing the
figure presented on the printing sector in the order in which the key
was operated and the ciphers in the orders to the right in case the
keys in the order to the right have not previously been operated.

The individual presentation of the type figures upon key depression,
except for the ciphers which were normally presented for printing,
required that in striking the keys, to give correct recording of the
items, the operation must be from right to left. That is, for example,
if the item to be added was $740.85, the operator would depress the (5)
key in the units cents column, the (8) key in the tens of cents column;
the cipher in the units dollars column would be omitted, the (4) key in
the tens of dollars, and the (7) key in the hundreds of dollars column
would be struck.

The printing hammers were provided with means for resetting after being
tripped in the recording action. This means is connected with the paper
shift-lever, so that as the paper was shifted or fed upward, ready
for recording the next item, the printing-hammers were all reset and
latched on their respective trigger latches, ready for a new item.

Fixed to the shaft 97, on which the printing-hammers are pivoted, is a
bail, marked 98, part of which is shown in the drawing, the horizontal
bar of which normally lies under and out of the way of the hammers as
they plunge forward in printing. And attached to the right-hand end of
the shaft 97, is a crank arm connected by a link to the paper-shift
hand-lever, which may be seen on the right in the photo reproduction of
the machine. This connection is arranged so that depressing the lever
causes the shaft 97 to rock the bail 98 rearward, thus picking up any
tripped printing-hammers and relatching them.

The totals had to be printed, as in the first-described Felt recorder,
by depressing a key corresponding in value to the figure showing on the
wheel in each order.

[Sidenote: _Felt principle of printing adopted by all manufacturers of
recorders_]

The principle involved in the individual hammer-blow, combined with
the ordinal type sector for recording in a recording-adder was new,
and was the feature that has made the adding-recording machine of
today possible, as is well in evidence by the presence of this
combination in all the recorders that have been made by the successful
manufacturers of listing or recording-adding and calculating machines.
Some manufacturers have substituted a vertical moving type bar for the
pivoted sector, but the scheme is the same, as the purpose is to get
the arrangement of the type in columnar order, and does not change the
fundamental features of the combination which furnished the practical
means for the individual type impression.


THE FELT TABULATOR

[Sidenote: _Wide paper carriage for tabulating_]

The next feature in the Art, that has served in the make-up of the
up-to-date recorders, was the wide paper-carriage. This feature will
probably be recognized by many as a means supplied for the recording of
columns of items in series on sheet-paper.

As will be noted, roll-paper in ribbon form had been used in all the
previously illustrated and described recorders. While the Ludlum
patent shows a carriage, it had no capacity for handling more than a
single column of numerical items. The carriage in the Ludlum machine
was a feature necessary to the typewriter construction and offered no
solution to the feature of tabulating.

[Illustration: Felt Tabulator]

[Sidenote: _The wide paper carriage machine_]

The first disclosure of the wide carriage feature for tabulating was
in a machine made by D. E. Felt in 1889, which he exhibited to the
U. S. Census Bureau at Washington, D. C., in 1890. The machine was
also exhibited at the World’s Fair in Chicago, in 1893, along with
other products in this line of the Felt & Tarrant Mfg. Co. A photo
reproduction of this machine is shown on opposite page.

The machine was left at the Census Bureau, where it was used for
several weeks, and was very much liked. Felt made a contract to furnish
ten machines of this type, and the machine was recommended for purchase
by G. K. Holmes, Special Agent of the Census Bureau, but like many
other government department requisitions, the purchase order was never
issued.

Although this feature is now found in all first-class recording-adders,
the recording machine Art was too new in 1890 for the new feature to be
appreciated, and was not pushed, as there seemed to be no demand for
the wide carriage then. On this account Felt delayed applying for a
patent on his invention until 1899.

[Sidenote: _Litigation on tabulator patents_]

In 1904 a license under the patent was granted the Burroughs Adding
Machine Co., but soon after the granting of the license another
manufacturer of recording-adders brought out a machine with a
wide carriage, which was the start of a series of long-drawn-out
infringement suits. The fact that Felt had delayed taking out his
patent formed the grounds on which the Court finally decided that Felt,
from lack of diligence in applying for a patent, had abandoned his
invention, which made it public property.

The tags which may be seen tied to the carriage of the machine are the
official tags used to identify it as a court exhibit during the long
term of years the suits were pending in litigation.

Outside of the tabulating scheme, the machine was in other respects the
same as the recorder just described as the roll-paper “Comptograph.”

[Sidenote: _“Cross Tabulating”_]

The paper, as may be noted, is held in a shiftable carriage and is
operated by two levers, one to feed the paper vertically and reset the
printing-hammers, while the other moved the carriage laterally for the
spacing of the columns of items or the cross-printing when desired.
Besides the lever action for shifting and paper-feeding, means were
provided on the right-hand end of the carriage for performing these
functions; one of these is the thumb knob which served to feed the
sheet of paper into the rolls; the other is a small lever which allows
the operator to shift the carriage by hand independent of the carriage
shift-lever.


THE THIRD FELT RECORDER

While the first lot of recording-adders manufactured by Felt were
wholly practical, as was well proved by the statements of those who
purchased them, it is easy to pick out features in their make-up that
today, when compared with the new highly-developed Art, would seem to
make them impractical.

The necessity of operating from right to left and the necessity of
printing the totals by key depression were features that, in view of
there being nothing better in those days, did not seem objectionable to
those who used them. They were features, however, that Felt overcame
and eliminated in the next lot of machines manufactured and placed on
the market in 1890.

[Illustration: FELT'S COMPTOGRAPH

One of the Early “Comptographs”]

This lot of machines, one hundred in number (a goodly number in those
days), were equipped with a special hand-knob in front on the left side
for automatically printing the totals, and with means by which the
ciphers were printed only on operation of the paper shift-lever, which
allowed the operator to depress the keys from left to right or any way
he pleased.

[Sidenote: _Felt recorder in “Engineering” of London, Eng._]

The best evidence as to what these machines looked like is to be found
in the reproduction on the opposite page of an illustration which
appeared in “Engineering” of London, in 1891.

It will be noted that the patent drawings of the Felt calculator are
also displayed. They were used to describe the adding mechanism of the
recorder.

The total printing device is shown and described in patent No.
465,255, while the patent for the printing of the ciphers by the hand
shift-lever was not applied for until 1904.

It may be argued, and argued true, that these two later features in
their generic application to the recording-adding machine Art were
anticipated by Burroughs in his invention herein previously described.
But, assuming that these features were operative features in the
Burroughs machine, they could not be claimed in combination with a
printing mechanism that was operative to give practical results and in
themselves did not make the recording-adder possible. Nor was the means
shown for recording the totals of use except with means for legible
recording.

[Sidenote: _Total recording a Felt combination_]

[Sidenote: _Legible listing of items and automatic recording of totals
first achieved by Felt_]

There is no desire to discredit what Burroughs did, but let the credit
for what Burroughs accomplished come into its own, in accordance with
the chronological order in which it may be proved that Burroughs
really produced a machine that had a practical and legible recording
mechanism. Then we will find that to produce such proof we must accept
the fact that in all the successful recording machines manufactured and
sold by the Burroughs Adding Machine Co., the printing type-sector, the
printing type-hammers and the overlapping hammer-triggers with their
broad functioning features forming a part of Felt’s invention, have
been used to produce legible recording, and that the combination of
practical total printing was dependent on Felt’s achievement.

[Illustration: Gottfried Wilhelm Leibnitz]

We might say that broadly Burroughs invented means that could be worked
in combination with the Felt printing scheme to automatically print the
totals, which is in evidence in all the practical machines put out by
the Burroughs Co.

But such a combination was first produced by Felt in 1890, and was not
produced by Burroughs until 1892.

As has been shown, Felt built his recording scheme into his key-driven
calculating machine, and added the paper shifting-lever to furnish the
power which was utilized finally for setting the printing-hammers and
tripping them for the ciphers.

Such a combination divided the work, but made a two-motion machine,
whereas the adding mechanism was designed on the one-motion principle.
Now the principle of the two-motion machine was old, very old. The
great Gottfried Leibnitz invented the first two-motion calculator in
1694. (See illustration on opposite page.)

The Leibnitz machine was a wonderful invention and there seems to be a
question as to its operativeness. As a feature of historic interest,
however, it created considerable commotion in scientific circles when
exhibited to the Royal Society of London.

[Illustration: Leibnitz Calculator, made in 1694

The First Two-Motion Machine Designed to Compute Multiplication by
Repeated Addition]

The first really practical machine of this type, however, was invented
by a Frenchman named Charles Xavier Thomas, in 1820, and has since
become known as the “Thomas Arithmometre.”

The Thomas machine is made and sold by a number of different foreign
manufacturers, and is used to a considerable extent in Europe and to a
limited extent in the United States.

[Sidenote: _The key-set principle more practical for recorders_]

But two-motion calculators, from Leibnitz down to date, have always
been constructed so that the primary or first action involved merely
the setting of the controlling devices and performed no function in the
supplying of power to operate the mechanism which does the adding. With
such machines the load was thrown on to the secondary action.

This, of course, made the primary action of setting, a very light
action, especially when keys came into use, and as there are several
key depressions to each secondary or crank action, it may be understood
that while the action of Felt’s printing or paper shift-lever was
light, the action of the keys which were called upon to perform most
of the work was much harder than it would have been if his adding
mechanism had been designed on the key-set crank-operated plan of
the regular two-motion machine such as illustrated in the Pottin or
Burroughs patents described.

Thus, when Burroughs applied the Felt recording principle to his
key-set crank-operated adding mechanism, he produced a type of
recording machine which proved to be more acceptable from an operative
standpoint than the recorder made by Felt; and yet the writer has read
testimonials given by those who had both the Felt key-driven recorder
and the Burroughs key-set crank-operated recorders, who claimed they
could see no advantage.

[Illustration]

[Illustration: From Drawings of Burroughs’ Patents Nos. 504,963 and
505,078]

Probably the best proof lies in the fact that Felt finally abandoned
the key-driven feature in his recorders, as may be noted from the
later-day “Comptograph.”


THE FIRST PRACTICAL BURROUGHS RECORDER

The first Burroughs patent to show the successful combination referred
to was No. 504,963, applied for May 5, 1892, and issued September 12,
1893. The printing scheme, however, while indicated in the said patent,
was applied for in a divisional patent, No. 505,078, issued on the same
date. Drawings from both these patents are shown on opposite page.

[Sidenote: _Description of first practical Burroughs recorder_]

[Illustration: Burroughs Recorder]

The new printing device, as will be noted, instead of operating at
the bottom of the machine, operates at the rear and prints the paper
against a roll mounted outside of the casing.

Outside of adopting the Felt method of printing, the general scheme
of construction used in the machine of the former-described Burroughs
patent was maintained, except that the levers D, used to drag the
denominational actuators down, were omitted, and a series of springs,
one for each actuator, was supplied to pull such levers down as are
released by key-depression when the common actuator drops under crank
action.

Thus the description previously given will suffice for a general
understanding of the mechanical functions of the adding mechanism
and the general scheme for the setting up of the type in these later
patents.

The construction of the type sectors, the printing-hammers and the
trigger-latches used to retain the hammers against the action of their
operating springs is best shown in the drawings of patent No. 505,078
on page 136. Fig. 1 shows the normal relation, while Fig. 2 illustrates
the same mechanism in the act of printing.

The type sector as shown in drawings of patent No. 505,078 is marked K,
while in the drawings of No. 504,963 it will be found marked 611ᵃ. They
are formed from a continuation of the denominational actuators for the
total register in the same manner that the type-wheel gear racks h, of
the previously described Burroughs patent were formed.

The type u, are arranged on movable blocks marked 618, which are shown
held in their retracted or normal position by springs 682, but when
pressure is brought to bear against these type blocks in a direction
outward from the sector, the spring 682 will give and the type blocks
will slide outward in the slots provided to guide their action.

The paper, as will be noted, is fed from a roll, up between the type
and the printing-roll 599, in the same manner as the paper of a
typewriter, and through the interposition of an ink-ribbon between the
type and the paper, the pressing of the type against the ink-ribbon,
paper and roll gives imprint.

The pressure brought to bear on the type is through the hammer-blow
of the printing-hammers 715, of which there is one for each ordinal
printing sector. These hammers are pivoted to the rod 701, and
are spring-actuated through the medium of the pin 741, the lever
716, and spring 780, which, combined with the cam-slot w, in the
printing-hammers, serve to force the printing-hammers into the position
shown in Fig. 2.

The printing-hammers are normally retracted and latched by a series of
trigger latches 117, through the latch-tooth b, which engages the lever
716 at v.

Each trigger-latch 117, is pivoted on the rod 700, and provided with
an overlapping lug as shown in Fig. 4. These overlapping lugs, like
those described on the trigger-latches in the Felt patent, serve as
an automatic means of filling in the ciphers in the same manner as
described in the Felt machine.

The means for tripping the overlapping trigger latches naturally
differed from the means shown in the Felt machine, as the Burroughs
machine was not key-driven.

A very ingenious means for the tripping of the trigger-latches is
shown, consisting of the dogs 718, and rock-frame 711, and tie-rods
703-704, which co-operate with a cam-shoulder y on the arm of the
printing-sectors, to remain neutral or to disengage the trigger-latches
through a reciprocating action, shown in dotted lines in Fig. 1, patent
No. 505,078.

The tripping action takes place at the end of the forward motion of the
actuating hand-crank through connections not shown in the drawings.

It may be understood that on account of the overlapping of the
trigger-latches of the printing-hammers that if, as described in
relation to the Felt recording-machine, one of the trigger-latches in
any order to the left of the units order should be tripped, it would
cause all the trigger-latches to the right to be also tripped, and the
printing-hammers thus released to spring forward, giving an individual
hammer-blow for each type impression.

Thus, if the five-hundred-dollar key should be depressed, only the
trigger latch in that order need be tripped. This is brought about
through the fact that normally the tripping-dogs 718 are held out
of tripping engagement by the cam surface y of the type-sector, as
the rock-frame in which the dogs are mounted is moved forward in its
tripping action. But as the hundred-dollar order type-sector has been
lifted through the setting of the (5) key in that order, it allows the
tripping-dog to engage the trigger-latch of that order, and through
the overlapping feature of the trigger-latches to trip and print the
ciphers to the right.

It will be noted that the application of the printing-hammers varied in
detail from that of Felt much the same as placing the latch on the gate
post instead of on the gate. In the generic principle, however, the
individual hammer-blow for each individual impression was maintained.

[Sidenote: _Date of use of first practical Burroughs recorder_]

There have been many conflicting statements made regarding the date of
the first Burroughs listing or recording machine, which is probably due
to the fact that the statements were not qualified by such terms as
“practically operative” or “legible recording.”

Dates given as that of the first Burroughs recording machine range from
1884 to 1892. In a book published by the Burroughs Co. in 1912, under
the title of the “Book of the Burroughs,” there was a statement that
the first practical machines were made in 1891.

[Illustration: From the February 1908 Issue of Office Appliances
Magazine]

H. B. Wyeth, at one time sales agent for the Burroughs Co., and
whose father was president of the company in 1891 and several years
thereafter, testified in court that the first sale of a Burroughs
recording machine was made about December, 1892. Corroboration of his
testimony is set forth in a Burroughs advertisement which appeared
in the February number of Office Appliances Magazine in 1908, a
reproduction of which is shown on the opposite page.

That Burroughs was experimenting as early as 1885 is no doubt correct;
and that in this respect he antidated Felt’s first attempt to produce a
recording-adder, is not questioned. But when it comes to the question
of who produced the first practical recording-adder, there is no room
for doubt in face of the evidence at hand.

[Illustration]




Introduction of the Modern Accounting Machine


As the reader has been carried along through the tangle of mechanical
efforts of the men who have racked their brains to produce means that
would relieve the burden of those who have to juggle with arithmetical
problems and masses of figures in the day’s accounting, there was one
phase of subject that has not been touched upon. While these inventors
were doing their best to benefit mankind and, without doubt, with the
thought of reaping a harvest for themselves, the public, who could have
been the prime beneficiary, did not hasten to avail themselves of the
opportunity.

[Sidenote: _Opposition to the use of machines for accounting_]

In the early days, when the key-driven calculator was marketed, and
later when the recording-adder was also placed on the market, the
efforts of the salesmen for each of these types of machines, in their
endeavor to interest possible purchasers, were met with anything but
enthusiasm. Of course, now and then a wide-awake businessman was
willing to be shown and would purchase, but ninety-nine out of the
hundred who really had use for a machine of either type could not at
that early date be awakened to the fact.

Although the calculator and the recording-adder are indispensable
factors in business today, and have served to improve the lot of the
bookkeeper and those employed in expert accounting in general, they met
with very strong opposition for the first few years from employers of
this class. It was strongly evident that the efforts of book-keepers
and counting-house clerks to prevent these machines entering their
department were inspired by the fear that it would displace their
services and interfere with their chance of a livelihood.

Again, men of this class, and even those in charge of large
departments, took the mere suggestion that they had use for a
calculator or recording-adder as an insult to their efficiency, and
would almost throw the salesman out. Others would very politely look
the machine over and tell the salesman what a wonderful machine it was,
but when asked to give the machine a trial, they would immediately back
up and say that they had absolutely no use for such a machine; whereas
possibly now the same department is using twenty-five to a hundred such
machines.

[Sidenote: _Banks more liberal in recognition_]

Of the two classes of machines, the recording, or listing machines, as
they are commonly called, although a later product, were the first to
sell in quantities that may be called large sales. This was probably
due to the fact that they were largely sold to the banks, who have
always been more liberal in recognizing the advantages of labor-saving
devices than any other class of business.

The presence of these machines in the bank also had a tendency to
influence business-men to install recorders where the key-driven
calculator would have given far greater results in quantity of work and
expense of operating. In these days, however, the average businessman
is alive to his requirements, and selects what is best suited to his
needs instead of being influenced by seeing a machine used by others
for an entirely different purpose. The theory of using the printed list
of items as a means of checking back has blown into a bubble and burst,
and the non-lister has come into its own, not but what there has always
been a good sale for these machines except for the first four years.

[Sidenote: _Improvements slow for first few years_]

On account of the years it took to educate business into the use of
these two types of accounting machines, and the fact that the sales
of both were small at first, there were few improvements for several
years, as improvements depend upon prosperity.

Such changes as have been made since were largely aimed at refinements,
but there are some very noteworthy features added to the performance of
both types of machines, which are explained and described in following
chapters, where the subject will be treated under the class of machines
they affect.

[Illustration]

[Illustration: The High-Speed Calculator]




The High-Speed Calculator


As previously stated, the calculating machine was old when Felt
improved the Art by combining the key-drive with a plurality of
co-operative orders of adding mechanism. The advantage in the machine
he produced existed in the great increase in rapid manipulation which
it offered over the older Art, especially in addition. To improve
upon Felt’s contribution to the Art of calculating machines from a
commercial standpoint demanded a combination that would give still
greater possibilities in rapid manipulation.

[Sidenote: _Felt improvements on Comptometer_]

The patent records show that Felt again came to the front and gave
to the public a new machine containing many new combinations of
highly-organized mechanism that produced the above-named result. The
patents showing these features are Nos. 762,520 and 762,521, the two
patents being divisional patents of the same machine.

Although there were several patents on key-driven calculators issued to
others and a key-driven calculator placed on the market, which was sold
to some extent, none of these calculators offered anything that would
increase the possibility of more rapid manipulation than was to be had
from Felt’s old Comptometer.

[Sidenote: _Scientific distribution of functions_]

There is one feature about the machine of these two divisional patents
which stands out very prominently to those acquainted with the fine
points of the physical laws of mechanics. It is a feature that was not
printed into the specifications. It may be found only in the time
allowed for the mechanical movements to take place, which shows that
theoretical reasoning was the foundation for the distribution of the
functions in the machine of these patents into increments of time, and
that the arrangement of mechanism was especially designed to carry
out this primary theoretical reasoning. While it is obvious that such
procedure must accompany successful invention of mechanism, it is
seldom that we find such fineness displayed as may be found in the
timing of the mechanical functions of the later Comptometer.

The force of the above statement may be realized by study of the
mechanical motions of the old Comptometer and then trying to improve
on them to attain greater speed of operation. Such a possibility would
depend on more rapid key-strokes.

According to the physical laws of force and motion, to attain greater
speed of action demanded a decrease in resistance. Thus, less key
resistance must be attained to increase speed of operation.

Felt probably knew from experience that lighter key action could not be
had by juggling with springs or by polished surfaces. He was also aware
of the infinitesimal space of time allotted to each function, as the
parts of the mechanism flew about in the merry dance they performed in
whirling the numeral wheels around while under the manipulation of an
expert operator. He couldn’t see the parts work--he could only theorize
when there was trouble; thus he alone knew the difficulties to be met
in attempting to make a more rapid calculator.

To describe the mechanism of the new machine from drawings of these
patents would leave the reader still in the dark. What was really
accomplished can best be understood by reference to the mechanical
action in the old Comptometer.

In order that the reader may understand the significance of what
was accomplished, let him consider this fact; that the key action
of the old “Comptometer” measured as high as eighty-six ounces to a
key depression, while in the new machine made under the two named
later patents the key depression was reduced to but twenty-two ounces
maximum, or a little over a fourth of the power required to operate the
keys of the old “Comptometer.”

[Sidenote: _Power consumed by old carrying method_]

Facts show that a very large part of the resistance met with in the key
depression of the old machine was caused by the high tension of the
springs which performed the carrying. This high tension was necessary
on account of the extremely small fraction of a second allowed for the
performance of their function of supplying the power that turned the
higher wheel in carrying.

By referring to the description of the inoperative features of the Hill
machine (page 25) a parallel example of the time for the carry of the
tens in the old Comptometer may be found, showing that but a ¹/₁₆₅ of a
second was the allowance.

The carrying means employed in the old Comptometer consisted of levers
with dogs or pawls hinged on their free ends, which co-acted with the
ten pins of the higher numeral wheels to ratchet them forward a step at
a time. The power for supplying such ratcheting action, in the delivery
of a carry, was produced in a spring attached to the carrying-lever to
actuate it.

[Sidenote: _Cam and lever carrying mechanism_]

The means used to produce the power in the carrying-lever actuating
springs, or best termed carrying springs, was through the turning of an
envolute cam attached to the lower order numeral wheels, which, acting
upon an arm of the carrying levers, forced them away from the wheels,
and thus tensioned the carrying springs. The cam and lever is best
shown in Fig. 7, page 130.

The timing of the delivery of the carry, as the numeral wheel passed
from nine to zero, was brought about by the high point of the cam
passing from under the arm of the carrying lever, which, when released,
allowed the carrying springs to act and ratchet the higher wheel
forward a tenth of a revolution.

This form of carrying action had a peculiarity of reaching a certain
set tension when three wheels were employed, so that for all the wheels
employed in greater numbers no higher tension was required and no lower
tension could be attained. Another feature about this type of transfer
device was the fact that to get the set tension as low as possible
required that at least eight-tenths of the rotation of the lower wheel
should be utilized in camming back the carrying lever or storing the
power for the carry. A decrease in this timing meant an increase in the
resistance offered in turning the lower wheel by the steeper incline
of the cam, and when the wheel in turn received a carry, the increase
of resistance increased the work of carrying, and so on by a geometric
ratio.

[Sidenote: _One-point carrying cam impossible_]

In a recent patent suit, a physical test was made as high as three
orders with a one-point cam; that is, a cam operating to store power
during a one-tenth rotation of the lower wheel (not an uncommon
combination as shown in patents that have been issued), and it was
found that by the time the third carrying was reached the springs
were so large and powerful that to turn the next wheel would require
a railway-coach spring, and that under the same ratio a fifty-four ton
hydraulic press would be required to depress the keys in the eighth
order.

The foregoing illustration of the idiosyncrasies of mechanical
construction offer a good example of why perpetual motion is not
possible, viz., that no mechanism was ever made that would not consume
a certain per cent of the power delivered to it, through friction
and inertia. Of course, expert knowledge of the physical laws of
mechanics allow of the application of force along the lines of least
resistance, and it is with this feature that the new improvements in
the Comptometer have to do.

[Sidenote: _Felt’s improved method of carrying_]

It would seem that the old carrying means could not be improved upon
under the circumstances, but Felt conceived a means which gave more
time for the storage of power for the carry and all kinds of time for
its delivery, which decreased the power required for carrying by a
very large per cent. The means he devised was a motor-type of carrying
mechanism that could receive and deliver power at the same time without
interference. Thus the full revolution of the lower wheel could be
utilized in storage and the same amount of time could be consumed in
delivery if necessary, but it was never required.

This tremendous reduction in power required to turn the higher wheel
in a carrying operation so decreased the resistance of turning the
numeral wheels that the former means used to control the wheels
during actuation was unsafe; that is, the old method of jabbing the
stop detent between the pins of the numeral wheel to stop it was not
dependable with the increased speed that the numeral wheels revolved,
under the reduced resistance.

Again, the feature of time was at issue. The wheels could be whirled
at tremendous speed or at a very slow speed. A sudden jab at a key
with the finger sent the numeral wheels kiting ahead of the rest of
the mechanism so that the detent could not be depended upon to enter
between the right pins, which would result in erroneous calculation.

In the new machine, we find that to overcome this unevenness of action,
Felt reversed the ratchet action of the denomination actuators, so
that no wheel action occurred on their down-stroke under the action
of the keys, but on the upstroke of the actuators the numeral wheels
were turned by the power of the actuator springs stored by the key
depression, thus giving an even set rotating action that could not be
forced and that could be controlled by a stop detent.

As the timing of this stop-action was coincident with the stopping of
the actuators on their upstroke, the actuator was used to perform this
function in combination with a detent device that could be released
from the wheel independent of the actuators to allow a carry to be
delivered.

[Sidenote: _Gauging and controlling prime actuation_]

A feature worthy of note connected with this change is displayed in
the method in which Felt overcame the timing of the stop action of the
actuators in the downward action they received from the keys, which
would have been as hard to control as it was to control the wheels
under direct key action.

[Sidenote: _Alternating stop scheme_]

The scheme he devised gave more than double the time to perform the
function of intercepting the lightning action with which the actuators
moved under a quick key-stroke. The scheme shows a dual alternating
stop-action constructed by the use of two stops acting at different
levels and co-acting alternately with five equi-spaced stop-shoulders
on the front end of the actuators, which were also arranged in
different levels.

The two stops were actuated by the keys in a similar manner to the
single stop which co-operated with the pins of the wheel in the old
“Comptometer,” except that the odd keys operated one stop while the
even keys operated the other.

Thus in the new “Comptometer” the (1) key acted to throw the higher
level stop into the path of the lowest stop-shoulder on the actuator,
and the (2) key acted to throw the lower level stop into the path of
the same stop-shoulder on the actuator. In the same manner the (3)
and (4) keys caused the odd and even stops to engage the next higher
stop-shoulder on the actuator and so on with the rest of the keys.

As the spacing was doubled by the use of but five stop-shoulders, the
stops were allowed double the time for entry between the stop-shoulders
plus the space that the pin occupied as compared with former method,
which was considerably more than double the time allowed for the same
function in the old machine.

Besides the redistribution of mechanical functions, another very
noteworthy feature is found in these patents which, in the specific
means disclosed, constituted another distribution of time for
mechanical action. This in the capacity of the machine was what has
become commercially known as the “Duplex” feature.

In the old “Comptometer” it was necessary to operate the keys
alternately, as a carry from one order to a higher order might be
taking place and thus be lost in the action of the higher order wheel
while rotating under key-action.

[Sidenote: _Multiplex key action_]

In the machine of the later patents the carry was delayed while the
higher-order wheel was under key-action. The construction shown
consisted of a latch operated by the actuators, which, when the
actuator was depressed, latched up the delivery end of the motor
carrying-device so that a carry due to take place at that time would
be intercepted until the actuator returned to normal again, at which
time the carrying motor device was again free to deliver the carry.
This feature allowed the striking of keys in several or all the orders
simultaneously, alternately, or any way the operator pleased, which was
a great improvement in speedy operativeness.

[Sidenote: _Control of the carry by the next higher actuator_]

While the genus of this elastic keyboard invention consisted of control
of the carry by the next higher actuator, the specie of the generic
feature shown was the delayed control. The first production of this
generic feature of control of the carry by the next higher actuator
that gave the elastic keyboard-action is shown in the two Felt patents.

It may be argued that this new keyboard feature was simultaneity of
key-action and that simultaneity of keyboard-action was old. True
it was old, but the flexible simultaneity was new and depended upon
individuality of ordinal control for its creation, and Felt created the
ordinal control that gave the flexible keyboard.

Simultaneity of key-action was old in key-driven cash registers; such
invention as had been disclosed in this line, however, would defeat
the usefulness of simultaneity in a key-driven calculator. The useful
feature of depressing keys in several orders at once in a key-driven
calculating machine lay only in the increased speed of manipulation
that it could offer.

[Sidenote: _Forced simultaneous key-action old_]

Now such simultaneous key-action as had been invented and used on
cash registers was not designed with the thought of increasing the
speed of manipulation in such machines. The simultaneity of the cash
register was designed to compel the operator to depress the keys,
which represented the amount of the purchase, exactly simultaneous;
otherwise, by manipulation the proper registration could be made to
show on the sight-register and a short amount on the total-register.
It was a device to keep the clerk or salesman straight and prevent
dishonesty.

[Sidenote: _Forced simultaneity applied to a calculator impossible_]

If you have ever watched an expert operator using a “Comptometer,”
try to imagine that operator hesitating to select a group of keys and
depressing them exactly simultaneously as one is compelled to do on one
of the key-driven cash registers. And then, on the other hand, if you
have ever seen a key-driven cash register operated, try to imagine its
being operated at the lightning speed at which the “Comptometer” is
operated.[4]

[4] In making this comparison, the reader should be careful not to
confuse the later key-set crank-driven type like that of Pottin
described in the preceding chapter. It was the old key-driven type of
cash register which contained the forced simultaneity of key-action.

It must be understood that the exact or forced simultaneity of the cash
register scheme, if applied to a calculating machine, would lock the
whole keyboard if one of any of a group of keys the operator wished to
strike was depressed ahead of the others, and would thus prevent the
rest of the group from being depressed until the return of the first
key.

[Sidenote: _Flexible simultaneity of key-action a Felt invention_]

It is within reason that a locking action of that character would
even defeat the speed of key-action that was possible on the old
“Comptometer,” since an operator could overlap the key strokes in
that machine to a certain extent; whereas the forced simultaneity of
the cash register, if applied to the “Comptometer,” would prevent any
overlapping or the depression of a second key until the first depressed
key returned.

The only simultaneity of key-action that could provide a means of
speeding up the old “Comptometer,” or any machine of its type, was a
means that would leave key-depression free as to matter of time; one
that would be perfectly flexible in group manipulation, offering a
complete fluidity of motion such as not to hinder the fingering of the
operator.

The purpose of the mechanical means employed to give simultaneity in
the cash register was to lock all the keys depressed together and lock
all others against depression until the former returned. The purpose
of mechanical means employed in the Felt patent was to give perfect
freedom of key-action, whereas formerly the key manipulation of the
old “Comptometer” was restricted in the freedom of key-action, to the
extent of being limited to seriatum action.

The above discussion has been somewhat elaborately detailed to offset
statements that simultaneity was old in the key-driven Art. There is
no question as to the cash register type of inflexible simultaneity of
action being old before Felt patented his flexible type of simultaneity
of key-action for a key-driven calculating machine; but any statement
intended to convey the idea that Felt’s contribution of the flexible
simultaneity of key-action to the Art was not new, must come from
ignorance of the facts or malice aforethought.

[Sidenote: _Duplex Comptometer_]

This flexible keyboard “Comptometer” was given the trade name of
“Duplex Comptometer;” the term “Duplex” meaning that two keys could be
depressed, as distinguished from the seriatum one at a time key-action
formerly required. The term, however, fell short of setting forth the
capacity of such action, as it was, in fact, not restricted to mere
duplex-action--it was really a multiplex key-action having no limit
except the lack of fingers on the part of the operator to depress the
keys.

The validity of these patents has been sustained in litigation. The
technical scope of the mere claims has been disputed, as patent claims
sometimes are; but the broad newness and importance of the practical
calculative capacity achieved is beyond dispute. The recent machine
called the “Burroughs Calculator” has multiplex key-action, but it did
nothing to advance the practical capacity of key-driven calculating
machines.

[Sidenote: _Introduction of full-stroke mechanism_]

The operation of key-driven machines has always been attended more or
less with a feeling that a key-stroke may not have been completed,
especially by a novice in operating. Recognition of the possibility of
errors occurring through incomplete key-strokes in key-driven adding
mechanism was first disclosed as early as 1872 in the Robjohn patent
(see page 36), in which a full-stroke device is shown co-acting with
the keys.

In the drawings it will be noted that for each key there is provided a
ratchet device co-operating with the key to compel a full-stroke. This
scheme, like other similar later attempts, was aimed at the prevention
of an error in the operation of adding mechanism, but as a means of
prevention of an error it was lacking, because unless the operator
noticed that the key had not returned the next key depressed would,
through the action of the rotor, pull the partly depressed key way
down until it was released, when it would rise again, possibly without
the knowledge of the operator. There still remained the fact that the
occurrence of the error was not made known to the operator until it was
too late to correct it.

[Sidenote: _Error signal keyboard_]

That Felt was interested in the solution of the problem for detection
and correction of the errors in key-strokes is shown in the several
patents issued to him on features pertaining to this subject. After
numerous experiments Felt came to the conclusion that it was futile
to lock a key in event of a partial stroke and that the solution lay
in the locking of the keys in the other orders from that in which
the error had been made, thus signaling the operator and compelling
correction before further manipulation could be accomplished.

Again we find, as with the simultaneity of key-action, that a question
may be raised as to the novelty of invention by those who wish to say
that there are full-stroke mechanisms in the key-driven cash register
Art that lock the rest of the keyboard. But the key-locks disclosed
in the cash register were directed to a continuity of stroke engroup,
as distinguished from the individualism necessary to the key-driven
calculator.

The mechanical means employed, of course, varied greatly from that
which would be of any value in the calculating machine Art, and the
theoretical scheme was aimed at a widely different result. Flexibility
was necessary.

[Sidenote: _Locking of the other orders by a short key-stroke_]

The feature sought by Felt for his calculator was a signal to the
operator that an error had been made--if an error should occur--and
to block the operation of any of the other orders until the error was
corrected. This he accomplished by causing all the other orders to be
locked against manipulation, through the occurrence of an error in a
key-stroke; thus preventing manipulation of another order until the
error was corrected.

[Sidenote: _Inactive keys locked during proper key-action in cash
register_]

Now it may be said that the locking of other orders was old in the
cash register; but let us analyze the scheme and action of both. The
depression of a key of the key-driven cash register immediately locked
all other keys not depressed, and retained such locking-action during
depression and until the complete return of such key-depression; thus
the keyboard was locked, error or no error.

[Sidenote: _Inactive keys not locked during proper key-action in
“Comptometer”_]

A correct depression of a key in Felt’s new invention, as applied to
key-driven calculators, does not lock the rest of the keys. In fact, no
key of Felt’s invention is locked until an error occurs.

The lock of the key-driven cash register is a lock that takes effect
without an error having occurred--one that is always present with
respect to the keys not depressed simultaneously, and a feature
designed to force simultaneity of group key-action to prevent, as
before explained, dishonesty.

The lock of the key-driven calculator inventions referred to are
in no way connected with simultaneous key-action--as in the cash
register--and never act to lock the other orders except when there is
an error in a key-stroke. As the writer has explained respecting the
simultaneous feature of the cash register, the locking of the other
orders in the cash register interfered with the flexibility of the
key-action and for that reason would be impossible in a key-driven
calculator, where rapid manipulation is dependent on flexibility.

The scheme of the new key-driven calculator inventions referred to,
were designed to allow perfect freedom of individual key-action and
to block such action only when an error in any individual key-stroke
should be made. There is nothing in common in the two schemes. The
time, purpose and mechanical means employed differ entirely.

[Sidenote: “_Controlled-key Comptometer_”]

This new idea of Felt’s is embodied in what is commercially known as
the “Controlled-key Duplex Comptometer.” The term “Controlled-key” was
coined to fit this broadly new combination, but a word coined to fit
the functions of a new mechanism is seldom enough to convey a complete
understanding of its true qualities.

Aside from the broad newness of the Felt “Controlled-key” feature
referred to, even the mechanical means for safeguarding the individual
key-action was new in its application as a full-stroke device. The
means employed operated directly on the accumulator mechanism, locking
it against registration until the error was corrected, which differed
greatly from the devices applied to the keys or actuators designed by
others to bring about a similar result. But the locking of all the
other orders of mechanism, through any key-action short of a full
stroke, as a signal or error, has no mechanical equivalent or simile in
the Art.




The Improved Recorder


[Sidenote: _The mass of recorder inventions patented_]

Since the general installation of the recording-adder by the banks,
the minds of “get-rich-quick” inventors have been turned toward this
type of machine. The result has been that a vast number of patents
on such machines were issued, a large proportion of which represent
worthless and impossible mechanism purported by their inventors to
contain improvements on the Art. Some of these patents on alleged
improvements describe and purport to contain features, that, if really
made operative in an operative machine, would be useful to the public.
But as inventions, they merely illustrate the conceptions of a new
and useful feature that can never be of use to anyone until put into
concrete operative form.

To describe these features would be useless, as they have not advanced
the Art; they merely act to retard its advancement through the patent
rights that are granted on the hatched-up inoperative devices or
mechanism purported to hold such features.

[Sidenote: _But few of the recorder patents of value_]

Of the vast number of patents issued, but few of the machines
represented therein have ever reached the market, and of these
machines, except those previously mentioned, there is little that
may be said respecting new elementary features that may be called an
advancement of the Art. It is to be expected, of course, that the
manufacturer of such machines will not hold the same opinion as the
writer on this subject. But the fact that the generic principles of
recording the items and totals were worked out before they even thought
of constructing such a machine leaves little chance for anything
but specific features of construction for them to make that may be
considered new.

[Sidenote: _Reserve invention as good insurance_]

Another feature to be considered in this line is that while these new
manufacturers were working out the “kinks” or fine adjustments, which
can only be determined after a considerable number of machines have
been put into service, the older manufacturers were working or had
worked out and held in reserve new improvements that were not obvious
to those new at the game.

It is quite common for manufacturers to have a reserved stock of
improved features to draw from. In fact, such a stock is sometimes
the best insurance they have against being run out of business by a
competitor who places a machine on the market to undersell them. Of
course, all manufacturers believe they purvey the best and advise the
public relative to this point in their advertisements.

[Sidenote: _Erroneous advertising_]

One manufacturer of a recording-adder, a much later invention than
either the Felt or Burroughs recorder, circulated some advertising
pamphlets once which contained a statement that their machine was the
first visible recorder. A reproduction of this pamphlet is shown on the
opposite page. The reader will at once recognize the error in such a
statement, as the first Felt recorder was a visible printer.

The statement seems extremely peculiar after paying tribute to Felt
as the pioneer in the Art of adding machines. One would suppose that
having knowledge enough of the Art to offer such tribute would have
left them better advised on the subject of visible recording.

[Illustration: Two Pages from Booklet Issued by Wales Adding Machine
Co.]

[Sidenote: _Error key_]

The first of the later improvements in the key-set crank-operated
recorder were made by Burroughs and consisted of the features which
formed a part of Burroughs patent No. 504,963 of 1893. One of these
features consisted of means provided in the shape of a special key
that when depressed would clear the key-setting, thus allowing of an
erroneous key-setting to be corrected by clearing and resetting the
correct item.

[Sidenote: _Sub-total_]

Another feature was provision for printing a total at any time without
clearing the machine, thus allowing printing of what may be called a
sub-total, while the grand total is carried on to be printed later.

[Sidenote: _Repeat key_]

The third feature consisted of means for repeated addition and
recording of the same item. The means provided consisted of a key,
which, if depressed after setting an item on the keys, would prevent
the keys from being cleared; thus by repeated operation of the
hand-crank the item set up would be printed and added repeatedly.

[Sidenote: _Locked keyboard_]

The next feature was one of construction, as it was designed to
overcome the possibility of the setting of two keys in the same order,
by locking all the other keys in that order. The invention was shown
applied to the Burroughs machine, but was applied for by Wm. H. Pike
Jr., and was issued January 13, 1898.

[Sidenote: _Quick paper return_]

In 1900 Felt perfected a quick paper return for his wide paper-carriage
and applied for a patent, which was issued March 11, 1902, the number
of which is 694,955. The feature was, that by operating a lever, it
served to return the paper after recording a column of items and
automatically shifted the carriage ready for the recording of another
column of items, thus facilitating speedy operation.

[Sidenote: _Paper stop_]

In March, 1902, a patent was allowed Felt on means to lock the
mechanism in a recorder when the paper was about to run out of the
rolls; a feature which, in tabulating, served as a check against the
paper running out of the rolls and prevented further operation until
the paper was shifted to commence a new column of items, thus insuring
the printing of each record on the paper which formerly depended upon
the vigilance of the operator.

[Sidenote: _Cross tabulating_]

The next feature in the recording machine Art which shows a new
operative feature, that may be considered an improvement, is
cross tabulating. It consisted of means for horizontal tabulating or
recording across a sheet of paper as well as in vertical columns. While
this feature was for special use, it served to broaden the usefulness
of the recorder in bringing together classified balances by dates with
cross-added totals, and many other similar uses. This feature was the
invention of D. E. Felt, who applied for a patent April 29, 1901, which
was issued October 21, 1902; the patent number is 711,407.

[Sidenote: _Item stop_]

Another special feature serving to broaden the usefulness of the
recording-adder was invented by Felt, and may be found in patent No.
780,272, applied for March 30, 1901, and issued January 17, 1905. This
feature was a device which controlled the printing of a predetermined
number of items which could be set by the operator, and which, when the
predetermined number had been printed, would lock the mechanism against
further action until the paper was shifted to print a new column.

[Sidenote: _Motor drive_]

Prior to May 9, 1901, there is no record of any recording-adder having
been operated by electric motor drive. But on that date Frank C. Rinche
applied for a patent showing such a combination with the recorder,
which became commercially known as the Universal Accountant. The
patent, No. 726,803, was issued April 28, 1903, and is the first of a
series issued to Rinche on various combinations of mechanical driving
connections.

[Sidenote: _Distinguishing marks for clear, totals and sub-totals_]

A feature common to recording of added columns of numerical items is
the distinguishing characters for clear, sub-totals and totals by the
use of letters, stars and other marks. The first patent on anything
of this nature that has come into general use was applied for June 9,
1903, by A. Macauley, and was issued June 12, 1906. This patent is No.
823,474, and shown connected with the Burroughs recorder to register
with a star when the first item is printed if the machine is clear and
when a total is printing. Provision was also made for printing an S
when a sub-total was printed.

[Sidenote: _Adding cut-out_]

The use of recording-adders is often applied when it is desired to
record dates along with tabulating added columns of recorded items.
Of course there is no use of adding the dates together, and again
if they were allowed to be added to the totals an erroneous total
of the columns added may result under certain conditions. Means for
automatically cutting out additions at certain positions of the paper
carriage in cross-line tabulating was devised by H. C. Peters, and a
patent showing such combination operative on the Burroughs recorder was
applied for by him May 12, 1904. The patent, No. 1,028,133, was issued
June 4, 1912.

[Sidenote: _Self-correcting keyboard_]

With the introduction of the key-set crank-operated feature on the Felt
Comptometer, the key action, like in the Burroughs recorder, became a
feature to be considered; but unlike the organism of the Burroughs,
the Felt construction allowed of the use of a self-correcting keyboard
without the possibility of error occurring from its use. This feature
is shown in a patent issued to Felt & Wetmore applied for December
27, 1904, and issued May 14, 1907. The patent number is 853,543,
and provides a means of correcting errors made in setting the keys
by merely depressing the proper key or keys, which will release any
previously set in the respective orders.

[Sidenote: _Split keyboard_]

In some classes of recording it is desirable to print more than one
column of items without shifting the paper carriage laterally. A means
providing for such an emergency is shown in patent No. 825,205, issued
to C. W. Gooch July 3, 1906. The patent was applied for December 2,
1905, and shows a means applicable to any order that may intercept the
printing of the ciphers in that order, and thereby the ciphers in all
other orders to the right from any key depression to the left of such
order. This made what has been generally known as the split keyboard,
but differs from that now in general use in that it was set to certain
orders and not selective at the will of the operator.

[Sidenote: _Dual action keyboard_]

With the coming of the motor-operated recording-adders, the extra time
allowed the operator, through being relieved of having to work the
crank back and forth, left a lapse of time until the motor finished
its cranking of the machine. In other words, there could be no gain in
the speed of operation because it took as much time for the motor to
operate the machine as it did by human power. In a patent granted to
McFarland, No. 895,664, applied for October 19, 1905, is shown a means
for utilizing the lapse of time which the operator was formerly obliged
to lose while waiting for the motor to finish its operation of cranking
the machine. It is shown in combination with the keyboard of the Pike
recorder and consists of a change that allows the keys for the next
item to be set while the motor is cranking the machine to print and add
the item previously set, thus utilizing the time formerly lost.

[Sidenote: _Non-add signal_]

In adding and recording columns of figures, it quite often happens that
it is desirable to print a number without adding it into the total,
which may be accomplished in general by depressing the non-add key or
knob, or what may be supplied for that purpose. These numbers, however,
were not provided with any means by which they could be distinguished
from those added into the total until Jesse G. Vincent conceived
the idea of printing a distinguishing mark beside them to designate
that they were mere numbers not added to the total. The means for
accomplishing this feature is shown in patent No. 1,043,883, applied
for September 24, 1906, and issued November 12, 1912.

[Sidenote: _Selective split keyboard_]

A new improvement in the split keyboard formerly devised by C. W. Gooch
is shown in a patent issued to Wetmore & Niemann applied to the Felt
“Comptograph.” This improvement consists of a selective device for
splitting the keyboard into four different combinations selective to
any combination. The patent was applied for April 24, 1907, and issued
February 2, 1915; the number is 1,127,332.

[Sidenote: _Selective printing cut-out_]

In some classes of recording it is desirable at times to cut out the
printing of some of the orders and in others the whole of the printing
mechanism. Mr. Fred A. Niemann patented a means for such a contingency.
The patent was applied for April 24, 1907, but was not issued until
March 9, 1920. The feature was shown applied to the Felt Comptograph
for tabulating or printing vertically a series of added and footed
columns of figures.

[Sidenote: _Grand totalizer_]

It is sometimes desirable to print the sum of all the totals of the
footed columns or what may be called a grand total. William E. Swalm,
in patent No. 885,202, applied for October 24, 1907, and issued April
21, 1908, shows how this feature may be accomplished on the Burroughs
recorder. It consisted of an extra series of accumulator wheels
that could be meshed with the regular accumulator wheels, and thus
receive actuation resulting in accumulation, the same as the regular
wheels. When, however, the regular wheels are zeroized in printing the
individual totals, the extra accumulator wheels are left out of mesh.
Thus the grand totals are accumulated. The printing of the grand total
is accomplished by the meshing of the grand total wheels with the
regular and the usual operation of taking a regular total. The regular
wheels, however, must be cleared first.

[Sidenote: _Alternate cross printing_]

The shuttle carriage, a means devised to print two columns of figures
by printing a number in one column and a sum in the other by alternate
action, was the conception of Clyde E. Gardner, and is shown applied to
the carriage of the Pike recorder in patent No. 1,052,811 of February
11, 1913. The patent was applied for September 24, 1908, and consists
of means for automatically shifting the carriage back and forth.

[Sidenote: _Determinate item signal_]

Another means than that invented by Felt to signal the operator when a
predetermined number of items have been recorded, consists of a bell,
which rings to notify the operator to that effect. This signal was
the invention of J. G. Vincent, and is shown in patent No. 968,005 of
August 23, 1910, and was applied for December 3, 1909, as an attachment
to the carriage of the Burroughs machine.

[Sidenote: _Subtraction by reverse action_]

Although subtraction has always been accomplished on this type of
machine as a means of correcting an error, it was always accomplished
on the Burroughs recorder by the use of what is generally known as
the complimental method, which, without special provision, is rather
objectionable. On the 22d of April, 1910, Wm. E. Swalm applied for a
patent which was issued June 4, which shows means connected with the
Burroughs machine that allowed subtraction to be made by the direct
method by setting the keys the same as for addition. The patent number
is 1,028,149.

[Sidenote: _Selective split for keyboard_]

A further improvement on the split keyboard feature is shown in a
patent issued to Fred A. Niemann in which is shown an individually
selective cipher cut-out that splits the keyboard into any combination
at the will of the operator. The said patent is No. 1,309,692,
applied for October 7, 1912, and issued July 15, 1919, and shows the
improvement in combination with the Felt “Comptograph.”

[Sidenote: _Rapid paper insert and ejector_]

In some classes of listing or tabulating it is an advantage to enter
the paper and eject it with a rapidity that will facilitate the
handling of a large number of sheets, such for instance as the usual
bank statements. In patent No. 1,208,375 F. C. Rinche shows how he
accomplished this feature on the Burroughs recorder. The patent was
applied for July 21, 1913, and issued December 12, 1916.

Of the named improvements, of course, all are designed to fit the
requirements of the machines they are shown as a part of in the
drawings of the patent. They are also claimed as adaptable to other
machines of the type, but some are so specific to the machine they
form an improvement on that they are not adaptable to other makes.
Again some give results on the machine they form a part of that was
accomplished in a different way in another make.

Most of the improvements named, however, are of such a nature that the
broad feature disclosed is adaptable to all makes if mechanism should
be specially designed to suit such machines that will function to give
the result.

[Illustration]




The Bookkeeping and Billing Machine


An outgrowth of the recording-machine Art is represented in a new type
of recording machine especially adapted to bookkeeping and the making
out of invoices or reports where typewriting combined with arithmetical
recording is necessary. This class of work demands a combination of the
typewriter with adding and multiplying mechanism, having a capacity for
printing the totals of either addition or multiplication.

[Sidenote: _Early Combinations_]

Several attempts have been made to combine the typewriter and
adding-recorder; and there have been combinations of multiplying and
recording. Another combination that has been used to some extent for
bookkeeping and billing is an adding attachment for typewriters, but
all these combinations were lacking in one feature or another of what
may be called a real bookkeeping machine and billing machine.

The combination of the typewriter and multiple-order keyboard
recording-adders was too cumbersome, and the means employed for
multiplication on such machines required too many manipulative motions
from the operator. In simple cases of multiplication as high as fifty
manipulative motions would be required to perform an example on such a
machine.

[Illustration: “Moon-Hopkins” Billing and Bookkeeping Machine]

The combination of multiplying mechanism, either direct or by repeated
stroke, with the multiple keyboard has been made, but without the
typewriting feature they do not serve as a real bookkeeping and billing
machine.

The combination of the typewriter and the adding attachment lacks
automatic means to print totals. The operator must read the totals and
print them with the typewriter. Multiplication on such a combination
is, of course, out of the question.

[Sidenote: _First Practical Combination_]

The culmination of the quest for a practical bookkeeping machine is
a peculiar one, as it was dependent upon the ten-key recorder, which
has never become as popular as the multiple-order keyboard on account
of its limited capacity. The simplicity of its keyboard, however,
lent to its combination with the typewriter, and the application of
direct multiplication removed a large per cent of the limitation
which formerly stood as an objection to this class of machine when
multiplication becomes necessary.

For the combination, which finally produced the desired result, we
must thank Mr. Hubert Hopkins, who is not only the patentee of such
a combination, but also the inventor of the first practical ten-key
recording-adder which has become commercially known as the “Dalton”
machine.

[Sidenote: _Moon-Hopkins Billing Machine_]

His bookkeeping machine is commercially known as the “Moon-Hopkins
Billing Machine.” See illustration on opposite page.

The term “Bookkeeping Machine” has been misused by applying it to
machines which only perform some of the functions of bookkeeping.

    The principle of “Napier’s Bones” may be easily
    explained by imagining ten rectangular slips of
    cardboard, each divided into nine squares. In the
    top squares of the slips the ten digits are written,
    and each slip contains in its nine squares the first
    nine multiples of the digit which appears in the top
    square. With the exception of the top square, every
    square is divided into parts by a diagonal, the units
    being written on one side and the tens on the other,
    so that when a multiple consists of two figures they
    are separated by the diagonal. Fig. 1 shows the slips
    corresponding to the numbers 2, 0, 8, 5, placed side
    by side in contact with one another, and next to them
    is placed another slip containing, in squares without
    diagonals, the first nine digits. The slips thus
    placed in contact give the multiples of the number
    2085, the digits in each parallelogram being added
    together; for example, corresponding to the number
    6 on the right-hand slip we have 0, 8 + 3, 0 + 4,
    2, 1, whence we find 0, 1, 5, 2, 1 as the digits,
    written backwards, of 6 x 2085. The use of the slips
    for the purpose of multiplication is now evident,
    thus, to multiply 2085 by 736 we take out in this
    manner the multiples corresponding to 6, 3, 7 and
    set down the digits as they are obtained, from right
    to left, shifting them back one place and adding up
    the columns as in ordinary multiplication, viz., the
    figures as written down are

        12510
        6255
      14595
      --------
      1534560

[Illustration: FIG. 1.]

[Illustration: FIG. 2. Napier’s Bones

From Napier Tercentenary Celebration Handbook]

    Napier’s rods or bones consist of ten oblong pieces
    of wood or other material with square ends. Each of
    the four faces of each rod contains multiples of one
    of the nine digits, and is similar to one of the
    slips just described, the first rod containing the
    multiples of 0, 1, 9, 8, the second of 0, 2, 9, 7,
    the third of 0, 3, 9, 6, the fourth of 0, 4, 9, 5,
    the fifth of 1, 2, 8, 7, the sixth of 1, 3, 8, 6,
    the seventh of 1, 4, 8, 5, the eighth of 2, 3, 7,
    6, the ninth of 2, 4, 7, 5, and the tenth of 3, 4,
    6, 5. Each rod, therefore, contains on two of its
    faces multiples of digits which are complementary to
    those on the other two faces; and the multiples of a
    digit and its complement are reversed in position.
    The arrangements of the numbers on the rods will
    be evident from fig. 2, which represents the four
    faces of the fifth bar. The set of ten rods is thus
    equivalent to four sets of slips as described above.

[Illustration]

[Illustration: From Drawings of Barbour Patent No. 130,404]

It is unnecessary to go into the history of the Hopkins Bookkeeping
Machine to show the evolution of the Art relative to this class of
machines, as the features that have made such a machine practical were
developed by Hopkins himself, and at the present date there is none to
dispute the title since his is the only machine having the required
combination referred to. The scheme used by Hopkins for multiplication
in his billing machine is, as stated, direct multiplication or that
of adding the multiples of digits directly to the accumulator numeral
wheels instead of pumping it into the accumulator wheels by repeated
addition of the digits as is more commonly used.

[Illustration: John Napier]

The direct method of multiplying is old, as a matter of fact, the first
mechanical means employed for multiplying worked by the direct method.
But its combination with recording and typewriter mechanism invented by
Hopkins was new.

[Sidenote: _Napier’s bones first direct multiplier_]

Napier, in 1620, laid the foundation of the mechanical method of direct
multiplication when he invented his multiplying bones. The scheme of
overlapping the ordinal places is shown in the diagonal lines used to
separate units from the tens in each multiple of the nine digits (see
illustration, page 179), thus providing a convenient means by which the
ordinal values may be added together.

[Sidenote: _First direct multiplying machine_]

The first attempt to set Napier’s scheme to mechanism that would add
and register the overlapping ordinal values was patented by E. D.
Barbour in 1872. See reproduction of patent drawings on opposite page.


THE BARBOUR MULTIPLIER

The accumulator mechanism of the Barbour machine, including the numeral
wheels and their devices for transferring the tens, is mounted in a
sliding carriage at the top of the machine (see Fig. 1), which may be
operated by the hand-knob.

[Sidenote: Description of Barbour Multiplier]

Extending through the bottom of the carriage are a series of pinions,
one for each ordinal numeral wheel, and connected thereto by a ratchet
and pawl action. The pinions are each so arranged as to be operative
with a gear rack beneath the carriage when the carriage is slid back
and forth.

Thus the wheels received action from one direction of the motion of the
carriage and remain idle during the movement in the other direction.
The degree of motion so received would, of course, depend upon the
number of teeth in the racks below encountered by the pinions.

The gear racks employed by Barbour were numerous, one being provided
for each multiple of the nine digits, arranged in groups constituting
nine sets mounted on the drums marked B (see Fig. 4). Each of these
sets contain nine mutilated gear racks, the arrangement of the teeth of
which serve as the multiples of the digit they represent.

The teeth of the racks representing the multiples of the digits were
arranged in groups of units and tens. For instance: 4 × 6 = 24, the
rack representing the multiple of 4 × 6 would have two gear teeth in
the tens place and four gear teeth in the units place, and likewise for
the eighty other combinations.

Adding the multiples of the digits by overlapping the orders was
accomplished by a very simple means, the arrangement of the racks being
such that as the carriage was moved from left to right the numeral
wheel pinions would move over the units rack teeth of a multiplying
rack of one order and the tens rack teeth of a multiplying rack in the
next lower order.

By close examination the reader will note from the drawings that the
lower one of the sets of multiplying gear racks shown on the drum B,
to the left in Fig. 4, is the series of one times the nine digits, the
next set or series of racks above are the multiplying racks for the
multiples of two, the lowest rack in that series having but two teeth,
the next higher rack four teeth, the next rack six and the next eight.

So far no multiple of two has amounted to more than a units ordinal
place, therefore these racks operate on a lower-order numeral wheel,
and are all placed to the right of the center on the drum B, but the
next rack above for adding the multiple of two times five requires that
one shall be added to a higher order, and is therefore placed on the
left side of the center of the drum.

Thus it will be noted that by reading the number of teeth on the right
of each rack as units and those on the left as tens, that running
anti-clockwise around the drum, each series of multiplying racks show
multiples of the digits from one to four, it being obvious that the
racks for adding the multiples of the higher digits are on the opposite
side of the drums.

From the layout of the racks it is also obvious that the starting or
normal position of the carriage would be with the numeral wheel pinions
of each order in the center of each drum, so that as the carriage is
moved to the right the units wheel will receive movement from the units
teeth of the rack on the units drum, while the tens wheel will receive
movement from the units teeth of the tens drum and the tens teeth of
the units drum, and so on with the higher wheels, as each numeral wheel
pinion except the units passes from the center of one drum to the
center of the next lower and engages such teeth as may be presented.

Each of the drums B are independently mounted on the pivot shaft C, and
are provided with the hand-operating setting-racks I and E, co-acting
with the gears R and D, to help in bringing the proper racks into
engageable positions with the pinions of the accumulator numeral or
total wheels.

The hand-knob G, Fig. 4, and the gears f, fast to a common shaft,
furnish a means for operating the whole series of drums when the right
multiple series of racks of each drum have been brought into position.

As an example of the operation of the Barbour calculator, let us assume
that 7894 is to be multiplied by 348. The first drum to the right would
be moved by its setting-racks until the series of multiplying racks for
adding the multiples of four are presented, the next higher drum to the
left would be set until the series of multiplying racks for adding the
multiples of nine were presented, the next higher drum would be set
for the multiples of eight, and the next higher drum, or the fourth to
the left, would be set for the multiples of seven. Then the hand-knob
G, first turned to register zero, may be shoved to the right, engaging
the pinions f with the gears D, and by turning the knob to register
(8), the first figure in the multiplier, the racks are then set ready
to move the numeral wheels to register as follows: The drum to the
right or the units drum has presented the multiplying rack for adding
the multiple of 8 × 4, thus it will present three teeth for the tens
wheel and two teeth for the units wheel. The tens drum presenting the
rack for adding the multiple of 8 × 9 will present seven teeth for the
hundreds wheel and two for the tens wheel. The hundreds drum presenting
the rack for adding the multiple of 8 × 8 will present six teeth for
the thousands wheel and four for the hundreds wheel.

[Illustration]

[Illustration: From Drawings of Bollee Patent No. 556,720]

The rack of the thousands drum representing the multiple of 8 × 7 will
present five teeth for the tens of thousands wheel and six for the
thousands wheel. Thus by sliding the carriage to the right one space,
the numeral wheel pinions will engage first the units teeth on one
drum, then the tens teeth on the next lower drum and cause the wheels
to register 63152. The operator, by turning the knob G to register (4),
the next figure of the multiplier, turns the drum so that a series of
multiplying racks representing multiples of 4 times each figure in the
multiplicand are presented, so that by sliding the carriage another
space to the right, the multiple of 4 × 7894 will be added to the
numeral wheels. The operator then turns the knob to register three and
moves the carriage one more space to the right, adding the multiple of
3 × 7894 to the wheels in the next higher ordinal series, resulting in
the answer of 2747112.

There are, of course, many questionable features about the construction
shown in the machine of the Barbour patent, but as a feature of
historic interest it is worthy of consideration, like many other
attempts in the early Art.


THE BOLLEE MULTIPLIER

Probably the first successful direct multiplying machine was made by
Leon Bollee, a Frenchman, who patented his invention in France in 1889.
A patent on the Bollee machine was applied for in this country and was
issued March 17, 1896, some of the drawings of which are reproduced on
the opposite page.

[Sidenote: _Description of Bollee Machine_]

Instead of using eighty-one multiplying gear racks for each order as
in the Barbour patent, Bollee used but two gear racks for each order;
one for adding the units and the other for adding the tens; these racks
operate vertically and are marked respectively Bb and Bc. (See Fig. 3.)

The racks are frictionally held against gravity in the permanent
framework of the machine, and are moved up and down by contact at each
end, received from above by bar Ie, and from below by pins of varying
length set in the movable plates Ab.

The bar Ie forms part of a reciprocating frame which moves vertically
and in which are slidably mounted the pin plates Ab. These plates are
what Bollee called his “mechanical multiplication tables.”

The arrangement of the pins and their lengths are such as to give
degrees of additive movement to the units and tens gear racks equal to
the multiplying racks in the Barbour multiplier.

The pin plates are moved by the hand-knobs Ab², and the plate shown in
Fig. 3 is positioned for multiples of nine.

The means for setting the multiples correspond to the index hand-knob
of the Barbour machine, and consists of the crank Am, which, when
operated, shifts the whole series of plates laterally. A graduated dial
serves the operator to set the multiple that the multiplicand, set by
the positioning of the plates, is to be multiplied by.

The accumulator mechanism is mounted in a reciprocating frame which
moves horizontally, causing the gears of the numeral wheels to engage
first the units racks on their upstroke under action of the pins, and
then the tens racks on their down-stroke under the action of the top
bar of the vertically moving frame, the downward motion, of course,
being regulated by the upward movement it receives from the pin that
forces it up.

As may be noted in Fig. 1, the multiplying plates are held in a
laterally movable carriage that is shifted through the turning of the
multiplier factor setting hand crank Am, by means of the rack and
pinion action. This gearing is such that each revolution moves the
multiplying plates under a higher or lower series of orders, thus
allowing the multiples of a higher or lower order series to be added in
the process of multiplication or subtracted in division, as the case
may be.

Although the Bollee machine is reputed to be a practical machine, as
is attested from the models on exhibit in the Museum of Des Arts and
Metiers of Paris in France, it was never manufactured and placed on the
market.

[Sidenote: _Bollee’s principle commercialized_]

Bollee’s principle has, however, been commercialized by a Swiss
manufacturer in a machine made and sold under the trade name of “The
Millionaire,” the U. S. patents of which were applied for and issued to
Steiger.

Hopkins constructed his multiplying mechanism on the Bollee scheme of
using stepped controlling plates for his reciprocating racks to give
the multiples of the digits, but the ingenious method of application
shown in the Hopkins patent drawings illustrates well the American
foresight of simplicity of manufacture.

During the past ten years there have been a large number of patents
applied for on mechanism containing the same general scheme as that
of Bollee and Steiger, but up to the present writing no machines with
direct multiplying mechanism have been commercialized except “The
Millionaire,” which is non-recording, and “Moon-Hopkins Bookkeeping
Machine.”




A Closing Word


As previously stated, it is impossible to describe or illustrate
the thousands of inventions that have been patented in the Art of
accounting machines, and some of the inventors may feel that the writer
has shown partiality. The subject of this book, however, has to do only
with the Art as it stands commercialized and those who are responsible
for its existence.

In the arguments to prove validity of contributions of vital importance
to the Art, many other patented machines have been used which really
have no bearing on the Art. But the writer was obliged to show their
defects, otherwise the misconception derived from articles written by
authors incompetent to judge would leave the public in error as to the
real truth relative to the Art of the modern accounting machines.

That all inventors deserve credit, even in the face of failure, is
without question. The hours, days, months, and sometimes years, given
up to the working out of any machine, intended to benefit mankind,
whether the result brings a return or not,--whether the invention
holds value, or no,--leaves a record that the world may benefit by, in
pointing out the errors or productive results.

If it were not for the ambitions and untiring efforts of men of this
type, who give heart and soul to the working out of intricate problems,
the world would not be as far advanced as it is today.

The writer has kept in close touch with the Art of calculating machines
since 1893, and made exhaustive research of it prior to that period.
There have been thousands of patents issued on machines of the class
herein set forth, but outside of the features reviewed there have been
no broadly new ones of practical importance that have as yet proved to
be of great value to the public. What is in the making, and what may
be developed later, is open to conjecture. It is a safe conjecture,
however, that in the present high state of the Art it will tax the
wits of high-class engineers to offer any substantial and broadly new
feature which will be heralded as a noticeable step in the Art. And
that, as in the past, thousands of mistakes, and impractical as well as
inoperative machines will be made and patented, to one that will hold
real value.

[Illustration]




Index to Subjects


    TYPES OF ANCIENT AND MODERN MACHINES                  Page
    General knowledge lacking                                        5
    Key-driven machine, first of the modern machines                 6
    Recording, the primary feature of adding machines that print     7
    Validity and priority of invention                               8
    Description of Pascal’s invention                               11
    Constructional features of the Pascal machine                   12
    Increased capacity of modern calculator                         13
    Patent office a repository of ineffectual efforts               14

    THE EARLY KEY-DRIVEN ART
    First attempt to use depressable keys for adding was
          made in America                                           17
    Description of Parmelee machine                                 18
    Foreign digit adders                                            18
    Single-digit adders lack capacity                               19
    Some early U. S. patents on single-digit adding machines        20
    Calculating machines in use abroad for centuries                21
    First key-driven machines no improvement to the Art             21
    Description of the Hill machine                                 22
    Hill machine at National Museum                                 25
    Inoperativeness of Hill machine                                 25
    High speed of key drive                                         26
    Camera slow compared with carry of the tens                     26
    Hill machine merely adding mechanism, incomplete as
          operative machine                                         29
    Chapin and Stark patents                                        29
    Description of Chapin machine                                   29
    Inoperativeness of Chapin machine                               30
    Description of Stark machine                                    33
    Inoperativeness of Stark machine                                37
    Nine keys common to a plurality of orders                       37
    Description of Robjohn machine                                  38
    First control for a carried numeral wheel                       41
    Description of Bouchet machine                                  42
    Bouchet machine marketed                                        43
    Misuse of the term “Calculating Machine”                        43
    Description of Spalding machine                                 47
    Prime actuation of a carried wheel impossible in the
          Spalding machine                                          49

    THE KEY-DRIVEN CALCULATOR
    Theory versus the concrete                                      50
    All but one of the generic elements solved                      51
    Originality of inventions                                       51
    A conception which led to the final solution                    52
    Evolution of an invention                                       55
    Trials of an inventor                                           55
    The first “Comptometer”                                         56
    Felt patent 371,496                                             56
    Description of Felt calculator                                  59
    Recapitulation of Art prior to Felt calculator                  60
    Why Hill failed to produce an operative machine                 61
    Idiosyncrasies of force and motion increased by use of keys     61
    Light construction a feature                                    62
    Operative features necessary                                    62
    Classification of the features contained in the early Art
          of key-driven machines                                    63
    Carrying mechanism of Felt’s calculator                         63
    Transfer devices                                                64
    Carrying mechanism versus mere transfer devices                 64
    Details of Felt carrying mechanism                              65
    Manufacture of the Felt calculator                              69
    Trade name of Felt calculator                                   70
    Felt calculator exhibit at National Museum                      70
    Significant proof of Felt’s claim of priority                   75
    Rules for operation an important factor of modern calculator    76

    EARLY EFFORTS IN THE RECORDING MACHINE ART
    First attempt to record arithmetical computation                79
    Description of Barbour machine                                  80
    Barbour machine not practical                                   81
    Description of Baldwin machine                                  82
    Baldwin’s printing mechanism                                    89
    First key-set crank-operated machine and first attempt to
          record the items in addition                              90
    Description of Pottin machine                                   91
    Early efforts of Wm. S. Burroughs                               95
    General scheme of Burroughs’ first inventions                   96
    Brief description of machine of early Burroughs’ patents        97
    All early arithmetical printing devices impractical            101
    Practical method for recording disclosed later                 102
    Inoperative features of early recording mechanism              105
    Adding mechanism attached to typewriter                        105
    Description of Ludlum machine                                  106
    Ludlum machine inoperative                                     108

    FIRST PRACTICAL RECORDERS
    Burroughs a bank clerk                                         111
    Felt interested in recorder Art                                111
    Felt’s first recording machine                                 113
    Felt recording mechanism combined with his calculating machine 113
    Description of Felt’s first recorder                           114
    First individualized type impression combined with
          printing sector                                          115
    First practical arithmetical recorder                          116
    The first sale of a recording adding machine on record         116
    Features of first practical recorder                           119
    Description of Felt’s second recorder                          120
    Felt principle of printing adopted by all manufacturers
          of recorders                                             124
    Wide paper carriage for tabulating                             124
    The wide paper carriage machine                                127
    Litigation on tabulator patents                                127
    “Cross Tabulating”                                             128
    Felt recorder in “Engineering” of London, England              131
    Total recording a Felt combination                             131
    Legible listing of items and automatic recording of totals
          first achieved by Felt                                   132
    The key-set principle more practical for recorders             135
    Description of first practical Burroughs recorder              137
    Date of use of first practical Burroughs recorder              140

    INTRODUCTION OF THE MODERN ACCOUNTING MACHINE
    Opposition to the use of machines for accounting               144
    Banks more liberal in recognition                              145
    Improvement slow for first few years                           146

    THE HIGH-SPEED CALCULATOR
    Felt improvements on Comptometer                               149
    Scientific distribution of functions                           150
    Power consumed by old carrying method                          151
    Cam and lever carrying mechanism                               152
    One-point carrying cam impossible                              153
    Felt’s improved method of carrying                             153
    Gauging and controlling prime actuation                        154
    Alternating stop scheme                                        155
    Multiplex key action                                           156
    Control of the carry by the next higher actuator               156
    Forced simultaneous key action old                             157
    Forced simultaneity applied to a calculator impossible         157
    Flexible simultaneity of key action a Felt invention           158
    Duplex Comptometer                                             159
    Introduction of full-stroke mechanism                          159
    Error signal keyboard                                          160
    Locking of the other orders by a short key-stroke              161
    Inactive keys locked during proper key-action in cash register 161
    Inactive keys not locked during proper key-action
          in “Comptometer”                                         161
    “Controlled-key Comptometer”                                   162
    The mass of recorder inventions patented                       163
    But few of the recorder patents of value                       163
    Reserve invention as good insurance                            164
    Erroneous advertising                                          164
    Error key                                                      166
    Sub-total                                                      166
    Repeat key                                                     166
    Locked keyboard                                                166
    Quick paper return                                             166
    Paper stop                                                     167
    Cross tabulating                                               167
    Item stop                                                      167
    Motor drive                                                    168
    Distinguishing marks for clear, totals, and sub-totals         168
    Adding cut-out                                                 168
    Self-correcting keyboard                                       169
    Split keyboard                                                 169
    Dual action keyboard                                           169
    Non-add signal                                                 170
    Selective split keyboard                                       170
    Selective printing cut-out                                     171
    Grand totalizer                                                171
    Alternate cross printing                                       171
    Determinate item signal                                        172
    Subtraction by reverse action                                  172
    Selective split for keyboard                                   172
    Rapid paper insert and ejector                                 172

    THE BOOKKEEPING AND BILLING MACHINE
    Early combinations                                             174
    First practical combination                                    177
    Moon-Hopkins Billing machine                                   177
    Napier’s Bones first direct multiplier                         181
    First direct multiplying machine                               181
    Description of Barbour Multiplier                              182
    Description of Bollee machine                                  188
    Bollee’s principle commercialized                              189

    A CLOSING WORD