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.