Part 3

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.

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.

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.

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.

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.

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.

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.”

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.

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.

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.