US1768806A - Card for numerical coding - Google Patents

Description

July l, 1930. c. H. TALLMADGE A u 1,768,806

r11 10, 1924 2 Sheets-Shawll :C mum/ July 1, 1930. c. H. TALLMADGE 1,768,806

CARD FOR NUMERICAL CODING v Filed April 10, 1924 2 Sheets-Sheet 2 JULY/0 021731660001000 ogolagago l/ /ll/ l 22Z|Z222 888 8868s amaaa 99s 992.99 99|99g Patented July l, 1930 UNITED STATES PMEN'Il OFFICE .CHARLES H. TALLMADGE F BUFFALO, NEW YORK, ASSIGNOR T0 LYNN A. WILLIAMS,

0F CHICAGO, ILLINOIS cam) For. N'UMERIGAL conm@ Application illed April 10,

My invention relates to a system or method of numerical coding by means of which all the easier natural numbers are equally distributed throughout nine or more great groups, and to the method of mechanically controlling the sorting of punched cards, the filing of documents, the routing of telephone communications and generally similar operations according to such novel method of coding which may be described as characterized by a left-hand lineup.

One of the objects of the invention is to increase the practical capacity of a qualitative field of two or more columns.

Another object of the invention is to increase the numbers in a code, thereby reducing the time and labor of punching the holes in record cards, and the likelihood to error in such punching.

A further object of the invention is to reduce the time and labor required in sorting the cards and the likelihood to pi or multilate them in the sorting operation.

Another object of the invent-ion is to permit the freer use of the smaller numbers of a given code without the carding and sorting handicap for such numbers, namely, the carding of such numbers by punching two, three or four holes instead of five and the sorting of them in two, three or four runnings instead of live.

A still further object is to distribute such small, better numbers equally among the conventional ten groups of a coding classification instead of concentrating them all in one group.

Other objects and advantages will appear in the further description of the invention.

Figure 1 is a vertical longitudinal section of a representative tabulating card sorting machine. Figure 2 is a vertical transverse section roughly on the line 2-2 of Figure 1 showing the adjustable s ring contact.

Figure 3 shows a perfldrated record card of the kind adapted to be sorted by the machine punched according to my invention.

Figures 4 to 13 inclusive show portions of cards similar to that shown in Figure 3.

One use of my method is perhaps most easily visualized by illustration from the me- 1924. serial No. 705,69?.

chanical card tabulating art, in which the data to be tabulated are transferred to punched cards, certain quantitative data in a field or fields of their own, to be summated, and other preliminary classificational data, such as ofiice number, or serial number, or account number, in other appropriate fields.

In a well considered classification scheme the several groups have important, less important and relatively unimportant classes of many grades, and eicient coding will attempt to assign the best number in each group to the most active class, the second-best numbers to classes of secondary importance, etc.

Obviously, a system that puts the several groups on a basis of approximate equalit as to best numbers is more desirable and e cient than one that crowds all the best numbers into one group to the exclusion of the others.

Ordinary righthand lineup grouping does the latter, when it does not altogether 1 nore the smaller numbers. Left-hand lineup grouping equalizes the distribution of the smaller numbers among the groups and utilizes all the best numbers, thereby increasing the economy of mechanical sorting and the eliciency of mechanical tabulation.

Before making any analytical summation of values or quantities, the cards are commonly run through a sorting device to separate them into groups and sub-groups according to the numerical code adopted for the particular classification in hand.

In some lines of heavy coding, typically represented by the Import Schedule in Foreign Trade, the present method separates and distiguishes its main groups according to the conventional numerical order, throwing numbers 1, 2, 3, 5, 6, 8, and 1o, 23, 99 and 123, 456 and 789, and, in fact all natural numbers below ten thousand into group 0, and preceding them by a punched cipher or ciphers in the case of a code with, say, five figures or columns. These ciphers must be sorted separately from the following significant figures by a separate running of the cards, one running for each column beginning at the left, whether occupied by a cipher or not..

The present invention is based on the acceptance of a novel organization of all the great groups of numbers by initial figures, whether the number is composed of one, two, three or more figures, thus throwing the above illustrative numbers into the following groups Stated in other words, the natural numerical series 1 to 99,999, when grouped according to the conventional numerical consecution of the prior art fallsinto ten groups, of which the first is 0, containing, let us say, all the one, tw0, three and four-figure numbers up to 9,999; the second contains all fivefigure numbers with initial 1. the third contains all five-figure numbers with initial 2. and the tenth all five-figure numbers with initial 9. l

But the 9,999 numbers of one, two, three or four figures in the first group of the old art are the very best of the whole series, and they may be equally divided into nine new groups of the present invention, of varying makeup as to actual numerical Value, with the nine digits as initial figures, much the 'same as the 90,000 five-figure numbers are now most naturally grouped according to their initial figures.

The left-hand lineup system of the present invention thus merely, but novelly and usefully,`fe xtends this initial-figure grouping to the numbers below, say 10,000, and combines such small numbers with the larger five-figure numbers already s0 grouped. It thereby establishes nine new great groups which share equally in the best (smaller) numbers, with practical advantages to be argued elsewhere ut which must be thoroughly visualized in practice to be fully appreciated.

With the inclusion of a tenth group, distinguished from the other nine by having an initial 0, followed by 0o (or 01) to 09, 00o (.001) to 099, 0000 (or 0001) to 0999, etc. (and the optional adoption` of 0, 0o, 000, 0000, etc. as class numbers), t-he full code of the new invention would consist of the following preferred numbers:

1st preferred, one-ti re 2nd preferred, twoure 8rd preferred, threegure 4th preferred, four-figure 1o numbers (1 to 0, inc.) 11o numbers 1,100 numbers 11,000 numbers a total, say, of preferred numbers 12,220 and, say, common numbers of five-figures 110,000

or a grand total (tive-co1- umn) capacity of 122,220 numbers a way as to adapt them to the rather inflexible code groups and the more or less varying and flexible classification groups. But few codes will have anything approaching this number of classes without expediently going into the sixth column, and with any code havinflr decidedly less demand for capacity, the better numbers would he the ones most likely to be used.

To consider more carefully the card-punching and tabulating art:

-In typical quantitative fields the conventional rule of right-hand lineup is, of course, necessary in order that units may be added to units, etc.; but in the preliminary punching and sorting of the cards into classification groups, the prior art practice of units under units is merely a conventional routine carried over along the line of least resistance by a natural but unsound analogy from the summation work. In such classificational fields there is commonly no end to be served by a summation; or, if a summation is made, it is merely for the purpose of checking or auditing, with little or no other significance attached to the numerical value of the total, and for such purposes the new left-hand lineup total is quite as desirable as the conventional kind.

Certain phases of the art, however, disclose serious handicaps in using the old righthand lineup in handling codes extending into three or more figures, the handicap becoming greater as the total capacity of the code increases to four or more figures. I have discovered that the adoption ofthe new lefthand lineup will remove these handicaps.

In practicing my invention in its simplest aspects, the several columns of the qualitative fields have no general units-tens-hundreds. significance, but the rule is first figure first column, meaning in each case the first one counting from the left.

Other things being equal, small numbers are better for code work than large ones, but in the prior art method of punching a best number, as 1, in a five-place code, it must be preceded by four ciphers, thus, ooool, and it must then typically be sorted five times, once for each column, beginning at the left. Recognizing this handicap, the post-office code ignores all numbers with less than five places, and codes the main office of the country, for example, New York main office, 20,000, punching it 2x and sorting it only twice, the

yX hole in this case substituting for the four ciphers following the 2 in the first column. Thus, in the old art it is seen that for convent-ional carding and sorting the number 20,000 is better than the simpler and gener ally easier number 2. Somewhat similarly, though not so wisely, Chicago is coded 10,100, punched 101x and sorted four times. Each dash in these cases indicates the saving of a sorting. In the prior art methods, the

numbers with four, three or two terminal ciphers thus become substantially preferred Y over the others in card work, since for such preferred numbers the holes punched and the cards run once in sorting may be reduced by 6o, 4o and 20 percent respectively of the rou tine requirements.

In thev left-hand lineup method, Chicago might be coded plain 1, and New York plaln 2, with seven other principal oliices coded from 3 to 9, and they would be carded 1x 2x etc., and each one sorted only twlce. Other principal ollices would be coded 10, 15, 20, 99, 10o, 1230, etc., and they would be carded as written, with thev X followlng the full number including any terminal clpher or ci hers in case the full number did not have the eld limit of figures. n

It does not seem practicable with the prior art methods, with the right-hand lineup and its fixed units-tens-hundreds limitations, to use the left-hand columns of a large field wholly independent of the right-hand co1- umns, butin practicing my invention 1t becomes quite feasible to do so. n

It is thus practicable, to have a total code capacity running into six 0r eight columns, if desirable for occasional minute sub-classifications, and still confine the main classes 1n more active use to the two, three or four columns at the left without handicapping the carding or the sorting of the more desirable small numbers. This cannot be done with the prior art code methods.

An extensive comparative analysis of the lsystem of my invention with the'old systems would be too bulky to include in this specilication, but such an analysis may be summarized as follows:

In the prior method, the Post Office code has a total capacity of 108,000 numbers with its two prefix series, against a corresponding total of 119,997 in the new art of my invention, and these two capacities may be analyzed item by item as follows:

First preferred: 9 two-run numbers in each art, with this difference: those of my new method are one-figure numbers while those of the old art are five-figured.

Second preferred: New method, 108 threerun numbers against old art 99. The new method numbers are twofigured^those of the old art are five-figured.

Third preferred: New method, 1080 fourrun, three-figure numbers against old art 972 four-run, five-figure numbers.

This completes all that are commonly considered preferred in the old art, and the three preferences may be summarized as follows:

New method 1197 numbers against old art 1080, an increase of 117 numbers or 11 percent. The new method numbers require 3465 figures for a single writing while the same number of numbers in the 01d art would require 5985 cent in figures over the same number .cent of the old art requirement.

But this new method has a further (sag1 fifth) set of preferred numbers against whic the old art has nothing to offer. They are 10,800 numbers which are written with four figures, though requiring five holes to be punched and also requiring, supericially, five runs in sorting. In the writing of these 10,800 numbers there is a saving of 20 perv of numbers in the 01d art.

In accounting and statistical work, it is no uncommon for some classes to be several hundred times as active as others, and where'the disparity in activity is marked, it is important that the severaleconomies in writing,

lcardinlg and sorting class numbers, especially the principal ones, should co-operate instead of conflict as they do in the prior art. This zo-operation is secured by the left-hand lineup.

It is thus apparent that while the smaller numbers of the natural series, sa the one, tw0, threeand four-figure num rs in the series 1 to 99,999, are the best and most easily handled in all ordinary respects, there is a heavy handicap in the prior art in the carding and sorting of these numbers as c0m- Eared with the carding and sorting of fivegure numberswhich ave two or more `terminal ciphers, and which employ the X skipbar as a substitute for such ci hers.

This handicap seems to have een responsible for 'the entire omission from the postoiice code of an7 of the natural numbers of four figures or ess, in spite of the obvious great advantages of such numbers .for general purposes.

In my invention, this very useful X skip is .employed even more freely than in the prlor art methods, but with wholly different reading and sorting signiicances, and, as one of the results, the heavy handicap above referred to is removed.

In the rior methods, the advantage `of thls use of comes partly from reducing the number of holes to be punched, but more from its sorting function by indicating that the unpunched columns at the right are really cipher columns. The punching of the extra holes thus saved would not be so laborious nor attended by so much delay and other wastes as the running of the cards for the extra sortings.

In the new method, the saving of punchings is even greater, but it arises in a different manner. The 01d methods cannot save the punching of holes for a variable number of places preceding the significant figures .of the columns actually needed for the use of small numbers like 1, 23 456, etc., in a five-place code. In practice, it must precede them bi something to fill the extra columns, and t is somethmg must be volitionvlil() ally graduated and laboriously inserted, step by step. It does not seem practicable to utilize the ski -bar for thisl purpose. But

by adopting the eft-hand lineup, it is practicable to write the necessary figures first and then follow them more mechanically, however variably, by the skip-bar which has its terminal limit automatically set for the first column of the next field, whether it be one, two 0r more columns away.

In the more important economy of sorting, as also in the lesser economy of punchin the new method saves by making availab e not merely an increased number of preferred ratin as well as an increased total capacity wit a given column limit, as will appear later, but by facilitating the assignment 0f such preferred ratings to active classes in greater ratio than the prior righthand lineup methods can do because such preferred numbers are distributed over all 4 the roups instead of being principally conether ignore the smaller numbers.

fine t0 one group. c

In a well considered classification scheme, each of the several groups has important, less important and relative] unimportant c lasses of many grades, and e cient coding .will attempt t0 assign the best number in each group t0 the most active class, the secondbest numbers t0 classes 0f secondary importance, etc. The total advanta ein punching and in sorting is the product 0 the single-use economy b the activity of the class.

Obvious y a system that puts the several basis of approximate equality as to best numbers is more desirable and eilicient than one that crowds all the best num- -bers into one grou t0 the exclusion of the other grou s. Or inary right-hand grouping does t e latter, when it does not alto- Leftand lineup grouping equalizes the distribution 0f the smaller numbers among the groups and permits a better utilizationof the best numbers in the scheme, thereby increasing the economy of mechanical sorting and the eiliciency of mechanical tabulation.

On the new method cards, 10,000- is written with the four ciphers, while on the oldmethod cards it is written 1X This might seem t0 be merely a transfer of the valuable skip-bar function from the number 10,000 t0 the number 1, which would be a worthy accomplis'hment in itself, since it removes the discrimination against the best 0f numbers, but further consideration reveals the further important fact: The left-hand lineup, with the transference of the skip-bar function of X from beinga substitute for two or more terminal ciphers to serving merely t0 complete the iield after the carding of a number of one, two, three 0r four ii ures, secures an increase of approximatel e even per cent in the working capacity 0 a tabulating card code, by using the left-hand columns entirely independent 0f the right-hand columns, with several distinct advantages besides the mere increase in capacity.

The use 0f smaller (preiix series auxiliary t0 the main series of natura numbers is perfectly unobj'ectionable from a card tabulating standpoint, and such series may sometimes be employed with advantage in heavy work, even when the total capacity thus secured is much in excess of the actual requirement, since in the new method it results in a marked increase of preferred numbers for writing, carding and sorting.'

With the increased capacity generally speaking about eleven percent) an especially with its very easy extension into six figures or more, obviously each group would utilize its one, two and three-figure numbers before going at all extensively into four iigures 0r more.

Thus, the Import Schedule, with only 83 classes in grou 7, which are now scattered from 70114 to 9361, uniformly five-figured and typically re uiring five runs for each class (card), mig t easil be coded entirely in small numbers, somew at as follows:

7 f 1 one-gure number, punched 2 holes, sorted 2 l'lllll t0 79 10 two-ligure, numbers, punched 3 holes, sorted 3 runs 771 72 three-gure numbers (with 28 unused threeligure numbers) punched with 4 holes but sorted in 3 runs, since the X in the fourth column would looksee as soon as the third sorting was accomplished provided there were no fourgure numbers in group 7.

sorted 3 times 810 thrd, 10,100 to 99,900, punched 101x etc. and sorted mes all of these preferred numbers being five-figured in writing.

The s stein wholly i nores the thousands of smal er numbers, wiliereas the left-hand system utilizes them in four classes of preference (comparably excluding the 0 and X series) thus:

9 iirs2ts, 1 to 9 inclusive, punched ixetc., sorting in runs 90 seconds, 1o to 19, 90 to 99, punched lox etc.,

sorting in 8 runs 900 thirds, 100 to 199, 90o to 999, punched 100x etc.,

sorting in 4 runs 9000 fourths, 1000 to 1999, 9000 to 9999, unched 1000x I etc., theoretically sorting in 5 runs, but in practice more often sorting in on y 4 runs.

It may be noted that all 0f these left-hand preferences are preferred in writing over any corresponding preferences in the righthand method. and that there is a very heavy fourth preference in the former which is entirely lacking in the latter.

After making all allowances for the adaptation of numerical groups to commodity groups, there would be an economy over the present method in the number of preferred classes, and in the nature of their preferences, with a still greater economy in the handling of cards punched for those preferred, more active classes.

Still further, the preferred numbers would be distributed rather evenly throu hout the schedule instead of being confine as now to group 0 (in Foreign Trade Schedules), where they cannot possibly be used to the' best advantage.

The 0 series of the postoice code, like its X series, respectively illustrated by 01000, Portsmouth, New Hampshire, and X1ooo, Honolulu, is an independent series of unit numbers of four digits each, with a special prefix symbol (well known in the coding art) employed in this case to bring units under units and thousands under thousands, and it is not merely an attempt to lise the natural numbers of the regular series beginning with 1. These prefixes must be written as well as punched and sorted.

In the foreign trade codes, on the other' hand, (some of) thesmaller numbers are frankly used, being written simply 10, 14, 19, A

for example, in the export code, and 101, 102, 191, etc. in the import code. It is only in carding and sorting them that they are given the prefatory ciphers (0010, 00101, etc.) and these ciphers clearly arise out of the technical carding and sorting necessities of the prior art, with its handicap of right-hand lineup in order to bring units under units.

In such carefully considered codes as those for postoilices and the foreign trades, the advantages thus typically inherent in small numbers cannot be fully realized. The conventional requirement of units under units handicaps the coding of any large series of oiiices or commodities because of the necessity of preceding the small number with one or more ciphers in punching, whether'or not the ciphers are recognized as part of the written code, and because the left-hand ciphers cannot be easily skipped, either in-carding or in sorting, as the right-hand ciphers may be, if desired.

Because of this, all large codes either ignore the small numbers or throw them into a single group by themselves, to the exclusion 4of other groups. This results in an embarrassment of wealth of small numbers in the 0 group, practically necessitating a great waste of the best numbers, since group 0 usually consists of onl a few hundred classes of commodities, w ile there are many hundreds of relatively small numbers in such code systems available for assignment,

and preferably to active or otherwise important classes.

As a further result, the other groups in such coding are precluded from participation in small-number assignment and are handicapped by an undesirable proportion- Group o, classes '101 to 9939 (carded and sorted 00101 to 09999) Group l, classes 10311 to 17801 Group 9, classes 90101 to 99999 But group 0 of the export code has only 138 classes and group 0 of the import code has only 289 classes, so that 861 small numbers (86%) of the one group and 9,710 small nurgbers (97%) of the other are whollyunuse In contradistinction the left-hand codes would be something like this:

Group 1, class 1 with 1 best number, classes 1o to 19 with 1o second best, classes 100 to 199 with 10o third best, classes 1000 to 1999 with 1000 next best, classes 10000 to 19999 etc. Similarly with the eight other regularly- *numbered groups:

classes 0o or 01) o o9 classes ooo 0r 001) to 099 Group would contain class o gli! desired ciasses 0000 or 0001) to 09999 etc. In all this left-hand system of codin however extended the number of figures, t ere is a vlsual continuity of numbers assigned in any group different from the ordinary numerical continuity, but perhaps even more valuable; and however extended the number of columns in a carded group, there would be a mechanical separation of main groups in the first running, a mechanical separation of sections in each oup in the second running, etc., so that botgrvisually and mechanically there would be a practical continuity or consecutiveness of the numbers of any group, or section, or sub-section, etc., each by itself, which would also permit the insertion of new numbers in their appropriate classification, quite beyond any economical practice of the prior art. To more concretely and tangibly illustrate my invention, an example of how the invention can actually be applied will be given.

Referring first to Figures 1 and 2 a brief description of the operation of a card sortf ing machine will be given. The machine shown is merely illustrative of any one of a number of machines which may be used in carrying out my invention. The illustrative machine is of the type used by the United States Census Bureau and shown and described in the atent to La Boiteaux No. 1,048,137 issued ebruary 19, 1924, so that a detailed description is not here necessary.

The cards 21 to be sorted are stacked in a hop r 22 and are advanced from the bottom of t e stack by any suitable feedin means suchv as a reciprocating plate23 an fed to the rollers 24 and guides 25. A spring pressed contact finger 26 is adjustably se cured to a transverse plate 27, the contact be i adapted to slide over one of the vertical co umns of the card as the card passes beneath it. The function of the machine is to sort the cards 21 into the pockets 28 corres nding to the perforations of the card in tb: column being sensed. A plurality of 'de leaves 29 pivoted or bent around a gled support- 31 have their posterior ends connected by strips 32 to the top of the pockets 28. The leaves 29 are arranged to be successively raised by a cam 33 vwhich is moved longitudinally to the right once during each cycle of the machine. This cam 33 is driven through an oscillating ratchet sector 34 which 1n turn receives oscillatory movement through a lost motion connection with the follower arm 35, the roller of which engages the slot of the cam 36. The cam 36 rotates at a constant speed, making one revolution for each cycle of the machine.

When the spring contact finger 26 passes through a perforation in the card, it swings the bell crank 37 counter-clockwise, thus releasing a dog 38 which is spring pressed so as to engage the ratchet sector 34 and is normally restrained by engagement with the lower end of the vertical arm of the bell crank 37. It will therefore be aparent that the position at which the ratc et sector is arrested. will depend upon the position of the perforation in the card assing beneath the spring contact finger. he cam 33 will also be limited in its longitudinal travel at a osition corresponding with the position o the perforation in the card. The leaves 29 will then be so positioned vertically relative to the last pair of rollers 24 that the card will be inserted between that articular pair of adjacent leaves which eflne a longitudinal path or ideway leadi to the corresponding poc et '28. Once t us introduced between a pair of adjacent leaves, the card is progressed along by the pusher arms 39 on an endless belt, until it dro sinto the pocket.

Althou h but a brief escription of this machine as been 'ven and many details have been omitted, 1t will be a parent that the sition (from the forwarcfedge of the cai-dgo of the perforation will contro the operation of the machine to car the particular record card to the cket which corres onds to the perforation 1n its column or lile ing sensed. It should-also be noted from Figure 2, that the spring contact member 26 may be adj ustably positioned so as to adapt it to sense the perfor-ations in any desired vertical column of the cards.

1'n Figure-3 I have shown a tabulating card 21 which is illustrative of the kind used in the post oflce money order auditing de artment. This card has columns in whic the various details of information such as the fee, serial number, issuing oilice number and file number ma be recorded by suitable punching. v To il ustrate one characteristic auditing operation, cards similar to that here shown representin the money orders cashed at a partlcular o ce must be sorted out according to the issuin oiiiee number. In Fi ures 4 to 13, only t e portions of the car which include the issuing oilice field are shown. Perforations are represented by a large black dot. The card 1n Figure 3 is punched to indicate that its correspondin money order was issued b oilice No. 1, whic we shall assume is the It will lbe seen that the perforations are made by beginning the unching at the left hand column of the liel irrespective of the number of digits in the oiiice desi ation. The 1 is punched in the extreme eft column and another perforation is made in the X position in the column second from the left. This is in contradistinction to the rior art method of punching where unc ings would be made for all of the zeros fore the first substantial digit (as indicated by the check marks on the drawings that is, the issuin oice designated 1-1 such designation ad been used in the old system of designations-would have been punched 00001; whereas by my method, the card will be punched 1x, thus saving three punching operations.

The table below gives exemplary desi ations for post oices in different cities w ich will be used concretely to illustrate how my invention may be carried out. For the purposes of this illustration, it will be assumed that there are 2400 cards to be sorted 800 in the Chicago (1)Ygrou or class punched 1x, 800 in the New ork 2) group punched 2x etc. The cards of course are not arran in groups but are shuled indiscriminately.

Assuming now that an bilico, say Detroit, has the above numbers of cards representing hicago Post Oiliceil lll money orders cashed from the particular cities as tabulated above, according to the old method of sortin generally employed, the cards would firste sorted accordin to the left hand column of the issuing office eld. It will be apparent that the determining perforations (accordin to the prior art as I have indicated by chec marks on the drawings) of each card occurs in the right hand file of this field and it will therefore be necessary to run all of the cards through the sorting machine five times. Since there are 2,400 cards, the machine necessarily will, in order to complete the sorting operation, virtually handle 12,000 cards,'that`is, handle each one of the 2,400 cards five times.

However, with the cards punched according to my invention and the sorting machine operated in accordance therewith, the operation will be as follows:

In the first sorting operation the contact finger will be set so that the machine will sort the cards according to the perforations of the left hand file of the issuin ofiice field. Of course all of the cards will ave to be passed through the machine in this first operation. It will be understood that for the second card sorting operation the linger is set to sense the second file from the left; for the third sorting operation it is set to sense perforation in the third file from the left; and so on progressively. v

In the second sorting operation the cards each having two punches or more have to be sorted. Since a card punchedfor even a single digit has to have two perforations, that is, a number perforationand an X perforation, this means that all of the cards have to be put through the second sorting operation. However, in the second sorting operation the cards which have only two punches, namely: 1X, 2X, 3X, 4X and 5X will be separated from the remaining cards and finally sorted. Therefore after the second sorting operation the two punch-cards need not again be handled by the sorting machine.

In the third sorting operation all cards having three or more punches are sorted, but the cards having only three perforations namely: 21X and 22X will be separated from the remaining cards and finally sorted. Similarly therefore the three punch-cards need not again be sorted.

In the fourth'punching operation the remaining cards, that is cards having four or more perforations are handled and the four punch-cards (in this instance cards perforated 224X) are separated from the other cards and finally sorted so that they do not have to be handled again.

The fifth sorting operation handles the remaining cards, which are cards having five perforations namely: 2241X, 22415 and 22417. The fifth sorting operation separates $1111 of these five punch-cards and finally sorts em. The total number of cards handled in each sorting operation is as follows:

-The total number of cards handled the old way would be 2400 times 5 or 12,000. This huge saving is not exaggerated. It is due in part to the fact that the cities that issue the greater number of money orders can be assigned the smallest designation numbers so that thecards punched with these small numbers are completely sorted in the second sorting operation (if they have one digit designation, but at the end of the third sorting operation if they have two digit designations). In the old method each card had to be run through the sorting machine five times regardless of the number of digits in the designation. 1

As previously pointed out, a great saving can sometimes be effected by looking to see if all the cards removed from one pocket have a perforation in the X osit-ion in the file according to the perforatlons of which the next sorting operation is to be made. Thus in the illustrative example all of the cards in the l, 3, 4 and 5 pockets at the end of the first sorting operation would have a perforation in the X position in the second file. This fact may be readil noted by the operator by aligning the stac of cards and running a wirethrough the X perforations in the second file or by merely holding the cards in alignment and observing that all of the cards are punched in the X position in their second files. This method of eliminating part of 'the sorting operation is well known and can be utilized in sorting according to the old method although it obviousl has greater advantages when employed with my invention.

My invention, as applied to tabulating card work relieves the art of the units-underunits handica of small numbers and makes mechanically feasible and practicable the distribution of the desirable numbers of a large code among the accepted classification groups with a fair approach of equality, thereb decreasing the mental andphysical wor involved in the handling of punched cards.

Advantages correspondingI in nature to 'those hereinbefore set forth in tabulating card work will also be easily secured in the generically related work of handling documents and communications which are similarly handicapped by a right-hand lineup in codlng or in operation.

8 lifzeaacds Inall this punched-card work, -the card is merely arconvenient tool or automatic means for controllin the operation of the sorting machine, whic operation in turn is merely the intermediate end in view preparatory to the final analytical tabulation of the quantitative data. v

In another closely relate'd aspect of the work, the documents from which the cards p have been punched sometimes needsorting into the same groupings as the cards to be filed for convenient reference, while in other lines the documents are sorted for filing in classified groups or in serial number order,

or both, without having had their data transferred to punched cards.

This sorting of documents may be accomlished either as a result of perforations ormed"in the documents themselves or b the operation of mechanism controlled directly b keys actuated in accordance with the num rs of the coded documents, and in this aspect the eneral advantages of an equal distribution of the smaller numbers of the natural 4seriesk are essentially the same as when applied to the punched-card work.

In still other lines which, for the ur oses of thisA speciiication, may be consi ere genericall similar, the mechanical setup or contro 'n means is neither for the purpose of tabulating or iilin but rather, as in telehone work, to estab ish the necessary sorting or routing of communications between senders and receivers, the latter being grouped accordin to localit into exchanges or the-like and t en indivi ually numbered or coded in a list whichy frequently embraces y thousands of names.

For the purpose of this specification, the names are analogous to classes in which cards are sorted for tabulation or to pockets into which documents are sorted for filing, in that the communications are classified or sorted and routed into receivin devices numbered according to the subscri rs on the codified list.

It would seem to make no diierence in the essence of the method whether the routing be determined by the senders in the automatic system, or by the operators at the central office in the manual system. The general advantages of the left-hand lineup coding are inherent in either system. The connections may be established more quickly and with less switching because of the greater number of small-number assignments, and the general public will more readily note and more easily remember the small numbers, with less liability to error in dialing or otherwise transmitting the code number for the controlling of the connections.

The punched card embodies a separate, permanent and tangible setup which may be emloyed to control the sorting at any convenient subsequent time, and the card punched by my` method may be used in the ordin way in connection with the ordinary car sorting machines, so that While the process of punching the card accordin to the novel grouping constitutes a metho by itself, in the narrow sense, the broaderA process consists in the novel makin of the tool and then using it to accomplish t e finally useful purose. The sortingof documents and the routing of tele hone communications, on 4the other hand, o not necessarily involve the making of any permanent setup, but do require the co-operation of more or less specially constructed devices, which receive and transmit the documents or the communications to their novell -grouped and novelly-selected resting p aces. The tem orary physical setu is thus prom tl.)1 fol owed by and merge with the actua distribution in a single process for each document or communication which corresponds to the broader rocess of making the card and using it, an in each case the advantages of the left-hand lineup are substantially the same.

Having thus described my invention, what I claim as new and desire to secure by Letters Patent of the United States is:

In a tabulating card sorting system, a gluralit of cards havin fields arranged in les an perforated to esignate numerical values rangin from one to three or more digits, adapte to be sorted by a machine for sorting the cards into ultimate classes, the total number of which runs into three or more digits, the sorting machine having a sensing element for progressively traversing the files of the fields of the respective cards for sorting the cards in res onse to the perforations therein, each card being perforated, beginlning with the first file sensed, for only the first si iicant digit and the digits to the ri ht t ereof in its corresponding numerical 'va ue with an x perforation in the le succeeding the file containing the last coded digit to the right.

In witness whereof, I hereunto subscribe my name this 5th da of A ril, 1924.

CHARLE H. ALLMADGE.

lll

cERTFICATE or coRREcTIoN. 9

Patent No. 1, 768,806. Granted July 1, 1.9.30, to

.CHARLES H. TALLMADGE.

f It is hereby certified that error appears in the printed siiecificaton of the above numbered'patent requiring correction as follows: Page 5, line 95, atten the word "Group" insert thel capital letter 0; and thatthe said Letters Patent shouldbe read with .this correction thereinthat the same may conform to the4 record of the case: in the.: Patent: Office.

Signed and aeal ed1t'his.26th. day of August, A. D. 1930.

' M. J. Moore, (Seal) Acting Commissioner of Patents.

Images