26t 



NA TURE 



[July i i, 1901 



must be so cumbersome as to leave the comptometer far behind 

 the more automatic arithmometers and so little better than head 

 and pencil work as to be a gain of doubtful value. 



When, further, he finds out that the inventor has evaded one of 

 the principal difficulties of arithmometer design, which relates 

 to the carrying of the tens, but which is due to the provision 

 that this operation must occupy the second half of the turn of 

 the handle and must, even then, be successive all down the row 

 so as to allow of the nearly simultaneous and overlapping oper- 

 ations on all the digits, in the comptometer it is not possible 

 where carryings come in to depress two keys simultaneously, for 

 in that case the carrying will fail. On the other hand, if the 

 keys are operated singly as many carryings as are necessary will 

 be accomplished. 



\Vhen, again, the arithmometrician, if I may so designate one 

 familiar with the use of the arithmometer, finds that the 

 comptometer, like the Income Tax man, can never subtract 

 anything (it can only add, and so apparently can never divide) 

 his despair is likely to be complete and he might well condemn 

 the machine as a toy. 



I will not go so far as to say that this exactly represented my 

 feeling when I began to prepare this notice, for I had known 

 the construction of the instrument for some years and was 

 generally familiar with it. However, I did feel that, from a 

 mechanician's point of view, it represented a retrograde step, and 

 it was only the knowledge that the comptometer was extensively 

 used in the United States, where appreciation of time-saving 

 appliances is more developed than here, that made me feel 

 that the comptometer must have advantages perhaps more than 

 sufficient to compensate for its operative deficiencies. 



The comptometer is a neat-looking instrument cased in 

 mahogany, occupying 14^ x 7^ inches on the table, and it is 

 four inches deep. On the upper surface there are, in the eight- 

 column machine, eight columns of spring-actuated number keys, 

 nine keys to each column. The lowest key of each column, 

 or rather the one nearest the operator, is marked in black i, 

 and these are called the i row, the next 2, and so on up 

 to 9. All the even keys are flat and the uneven concave, .so 

 the operator knows at once, without looking, if his finger has 

 got one row too high or too low. At the end next the operator 

 is a row of nine number wheels, or one more than the number 

 of columns, on one axle, seen through windows, so that only one 

 figure on each can be read. This is called the register. The 

 axle terminates outside on the right in a milled head, and below 

 this there is a liberator handle. If the operator finds any figures 

 on the result wheels that he does not want he presses the liberator 

 handle with one finger and begins to turn the milled head. He 

 then turns this as far as it will go, when nine o's will appear on the 

 number wheels. The machine is now ready to begin. If any 

 key is pressed down the figure shown in black on that key will 

 immediately appear on the corresponding number wheel below. 

 If it or any other key in the same column is pressed, the figure on 

 it will at once be added to the figure already on the number 

 wheel. If the result is more than 9, i will be carried to the next 

 number wheel to the left. If that should happen to be already 

 9, one will be carried on again and it will become o. If all the 

 figures are 9 and i is added to any one, then it and all to the left 

 will immediately become o. The action is almost instantaneous, 

 but not quite, as each number wheel on becoming 9 leaves a trap 

 set which it lets ofl' on becoming o. The trap then adds I to the 

 next number wheel to the left. If this is 9 the same thing 

 happens again, and so on across the machine as far as 9's happen 

 to extend ; so the action is really successive and the wave of 

 motion can just be detected if it is looked for. 



Any key instantly returns to its place under the action of a 

 spring when the finger is removed. The necessary movement of 

 the I keys is \ inch, while for the 9 keys ; inch is required with 

 intermediate movement for intermediate figures. The pressure 

 required is moderate, but more than is necessary for a type- 

 writer. The rate of striking the keys may become, with 

 practice, very great, so that, though numerous strokes are 

 required in a multiplication, the result may nevertheless be 

 found very quickly. Judging by the time that is stated to 

 be necessary for working certain examples, a rate of six or seven 

 strokes a second is certainly attainable, in fact, with but little 

 practice I find this to be possible and that the machine- works 

 correctly at this rate. 



The question will naturally arise here whether there is any 

 fear of overshooting by the wheels of the register, as they are 

 clearly set into very rapid rotation and have to be suddenly and 



exactly stopped. Various methods of stopping number wheels 

 are in use in arithmometers — spring clicks, cams like the 

 Geneva stop in clockwork, and a mere brake ; the method used 

 here is more direct and positive than any of these, for the key 

 at the end of its depression operates a long light lever which 

 brings a rigid stop between two pins on the number wheel of 

 the register, locking it absolutely and ensuring its stopping in the 

 correct position. The driving forward of the number wheels by 

 the keys is effected by a series of long light levers, each operated 

 by any one of the keys of one column. The 9 key is near the 

 fulcrum end, while the i key is near the number wheel end and 

 the others are in intermediate positions. A toothed arc at the end 

 of each lever gears with a corresponding pinion on the common 

 axis of the number wheels, and each of these pinions drives 

 round its number wheel by a ratchet and pawl. Each number 

 wheel in moving from o to 9 raises a light lever by means of a 

 cam to its highest position, which it lets drop on completing its 

 turn to o again. The lever in its descent moves on the next 

 wheel to the left one tooth. If, therefore, the key of that wheel 

 is being depressed at the same time, the carrying trap will not 

 move it an extra tooth, but will merely join with its operating 

 lever in moving it through one unit of movement, and the carry- 

 ing will be lost. 



To the left of each i key is a little push, which may be 

 pressed with one finger when any key in that column is being 

 depressed. This push throws the carrying trap out of gear with 

 the next number wheel, so that no carrying can take place. This 

 enables the operator to alter any figure in the result, or to bring 

 it to o by adding to it the necessary number .without, at the 

 same time, changing any other figure to the left. They are also 

 used in some special operations. 



I have now probably written enough to enable any one in- 

 terested in these machines to understand what the comptometer 

 is like and also its mode of operation. The next thing is to 

 explain how a machine that can only add, and only do that one 

 figure at a time, may nevertheless be used for performing all the 

 ordinary arithmetical operations, such as any arithmometer will 

 perform. 



Addition needs no more explanation. The speed merely de- 

 pends on the rate at which an operator can read the columns of 

 figures and get his fingers on to the right keys. A mere dab at 

 the key such as is desirable with a typewriter is not appro- 

 priate here, as the key must be pressed right down to its stop, 

 otherwise it may add a number less than that printed in black 

 upon its head. To acquire the proper stroke, high speed and 

 certainty of getting on to the right keys evidently requires 

 practice ; it would be interesting to see a really skilled operator 

 at work. 



In most arithmometers subtraction is effected (this is most 

 generally wanted for the purpose of division) by turning the 

 number wheels in the reverse direction, when the carrying acts 

 in the reverse direction also. It is merely addition backwards. 

 There is, however, a method of in effect subtracting on a 

 machine which, like the comptometer, does not admit of back- 

 ward motion. It is to add the arithmetical complement. This, 

 for instance, has been used in some operations in Mr. Edmond- 

 son's circular machine. If you wish to subtract, say, 7, you have 

 merely to add 3 and prevent the machine from carrying with the 

 push. If you wish to subtract, say, 29, you have merely to add 

 71 and prevent the second figure from carrying. Similarly, to 

 subtract, say, 234567S9, it is merely necessary to add 7654321 1, 

 each digit to be added being 9 - the one to be subtracted 

 except the last operative digit, which must be 10 -the one to 

 be subtracted, or i more than in the case of the others. If the 

 arithmetrical complement had to be found by the operator the 

 machine would not be of much use, but it has not. Every key 

 has a smaller figure in red upon it, which is 9 - the black 

 figure on the key. All that is necessary, therefore, in subtraction 

 is to work with the red figures, bearing in mind only that the last 

 operative figure to the right must be taken on the next key 

 above, and that the push belonging to the last figure on the left 

 must also be used to prevent carrying improperly. 



Multiplication of any number by another of one digit is, of 

 course, simple enough. To multiply, for instance, 37921 by 7 the 

 series of keys corresponding to the number 37921 are each struck 

 seven times, or else working on the 7 row the key farthest to the 

 right is struck once, the next to the left twice, the next nine 

 times, and so on. Either operation will produce the right 

 answer, but the second one is preferable because, having put the 

 finger on the last key of the seven row, there is no more occasion 



NO. 1654, VOL. 64] 



