Appendix B: Method of Tabulating Census Results

Next Selection Previous Selection

APPENDIX B.
METHOD OF TABULATING CENSUS RESULTS.

In previous sections of this Report, reference has been made to the method of tabulating the results of the Census of England and Wales, 1911. The following description of the new system adopted may be of interest to those engaged upon similar work in the various parts of the British Empire.

Owing to extension of the subjects of inquiry, and the demand for greater detail in the published results, the increased work consequently thrown upon the Census Office necessitated some alteration in the system of tabulation adopted at previous Censuses. After investigating the possibilities of the mechanical systems applicable to such an undertaking it was decided to adopt the Hollerith system. This system had been used by the Census authorities in the United States of America for the Censuses taken in the years 1890 and 1900. Without attempting any technical description it may be said that the general principle governing the operation of the machines employed in this system is the use of an electric current transmitted through wire brushes arranged to penetrate perforations in cards. The necessary cards and machines were supplied, and the latter were maintained, under contract, by the British Tabulating Machine Company, Norfolk Street, London, W.C. These machines were of four kinds: (1) Punching Machines for perforating cards; (2) Automatic Sorting Machines; (3) Automatic Counting Machines; and (4) Automatic Tabulating Machines. The cards were of three kinds: (1) Population Card ; (2) Personal Card; and (3) Fertility Card, facsimile reproductions of which will be found below. (Figs. 1, 2 and 3.)

It will be observed that the cards contain vertical columns of figures divided by lines into "fields." The number of columns in a "field" depends upon the varying number of items which have to be recorded; thus, a two-column field will provide for recording all numbers from 0 to 99, a three-column field will provide for recording all numbers up to 999, and so on. All the information to be recorded on the card has to be reduced to numerical form, and such of the answers to the Census inquiries as were not in this form had to be "coded," according to schemes prepared for this purpose.

The Punching Machines are of two kinds: (a) Key Punches ; (b) Gang Punches. The Key Punches are fitted with a keyboard similar to a typewriter. On a card being placed in the machine, the depression of a numbered key punches a hole in the card in the position assigned to the number indicated on the key. As the card is perforated it is automatically moved to the left, thus bringing the next column of figures underneath the knives of the punch. The Gang Punches are used for perforating packs of cards with holes common to all. These punches were utilised in punching the identification particulars, such as the county, urban or rural area and enumeration district in which a person was enumerated. A pack containing from 15 to 20 cards was placed in a gang punch at one time, and by the depression of a lever the various knives, which had been previously fixed in their appropriate positions, were forced through the cards.

The Sorting Machines are so arranged as to divide the cards, as perforated in any one column, into 11 groups. From a pack of cards placed in the machine each card is automatically removed from the face and carried past the brush. The position of the punched hole in the column on which the brush is operating determines the particular box into which the card is eventually sorted. By this means the cards can be arranged in numerical order of the significant numbers in any "field"; the number of times the cards require to be passed through the machine for the purpose being determined by the number of columns in the "field." The full capacity of the machine is 15,000 cards per hour.

The Automatic Counting Machines were required for counting the number of cards previously sorted, and also for recording the total number in combination with the numbers in sub-divisions of the main group of cards. In other words, the counting machine had a power of selection by combining its functions with. those of the sorting machine. The machines in use were of different capacity. The largest, with three rows of counters, was provided with a distinct counter for each position in which a hole could be punched in each. of three columns of the card, thus allowing a total to be obtained of the number of cards punched in each position in each of the three columns. In addition, three counters were provided for recording various subtotals and a grand total of cards counted. Other machines had only two rows of counters, and the simplest machines had only one, row of counters. These were used for obtaining results which only necessitated counting on two columns or one column at a time. As a practical example of the working of the machine, the operation of counting (a) the numbers employed in any one occupation ; (b) the numbers of persons at various ages in such occupation ; and (c) the numbers returned as employers, workers, &.c. (status), may be briefly described.

By means of the Sorting Machine the cards were sorted into numerical order of the occupation code. If the cards punched "812" (Paper Stainers) in the occupation field are selected for counting, one passage of the cards through the Counting Machine will furnish a result showing the total number of Paper Stainers, the numbers at varying ages, and the numbers of Employers, Workers, &c., the actual result being recorded as follows:—

Sample count of Paper Stainers by Age

For this particular count a machine provided with three rows of counters was used. One row of counters was reserved for counting the "Status" and two rows for "Age Groups"; 11 counters and a total counter were provided in each row. The first in each row was reserved for counting all cards punched "X," the second for cards punched "0," the third for cards punched. "1," and so on. A plug board enabled the counting brushes opposite to the "Age Groups" and "Status" columns of the card to be brought into operation to the exclusion of all others. The passage of a card punched "0" (= age 10) in the first of the age group columns operated the "0" counter in the row reserved for that column. At the same time a counter in the row reserved for recording Status was operated in the same way. Every card passed through the machine recorded 1 on the total counter, thus giving a figure to which those recorded on the counters reserved for Status or Age Groups must cast. The full capacity of these machines is 15,000 cards per hour, but the speed is always determined by the frequency with which results have to be recorded.

The Automatic Tabulating Machines were supplied for tabulating the Population and the Fertility results. In the Personal Card there were no fields which were additive, but in the two other kinds of cards used several of the fields related to records for which an aggregate was required. All such results could have been produced by the Counting Machines provided the cards had been sorted so as to arrange together those with holes punched in similar positions. The value of the hole and the total cards so punched being known, the sum of the values of the punched holes could be obtained. The use of the Tabulating Machine obviated the necessity of sorting on any field which was additive and the subsequent multiplication of the total number of the cards by the value of any particular punched hole.

The Tabulating Machines varied in adding capacity. One was capable of adding the results from three fields and others from two only. On these latter machines, the place of the third adding counter was taken by- a card-counting device. The advantage of such a machine is obvious, if Fig. 3 is referred to. On this card there were two additive fields (1) "Total children"; (2) "Dead children". When a card punched "05" in the "Total children" field and "2" in the "Dead children" field was passed through the machine, the result on each of the three counters was as follows:—

Cards. Children.
Total. Dead.
1 5 2

A second card, punched "14" in the "Total children" field and "5" in the "Dead children" field, passed through the machine, gave the following combined result:—

Cards. Children.
Total. Dead.
2 19 7

The ultimate result of the passing of a pack of Fertility Cards through the machine gave: (a) The total number of married couples (one card representing a married couple) ; (b) the total children born to such couples; (c) of such total children born, the number that have died.

The machine provided with three adding counters was used for adding the totals of Inhabited Dwellings, Males and Females enumerated in groups of enumeration areas. (Population Card, Fig. 1.)

The full capacity of these machines is 9,000 cards per hour, but the frequency with which results had to be recorded had a considerable effect upon the speed with which any particular series of cards could be dealt with.

The full number of machines of various kinds actually in use at one time was as follows:—

Key Punches 68
Gang Punches 8
Sorting Machines 15
Counting Machines 8
Tabulating Machines 2

The first step in the organisation of the new system of tabulating the results was the preparation of the various codes, all of which were arranged to facilitate the sorting processes. A number of clerks were then trained in the important work of coding the replies on the Census schedules and in adding the code numbers to the various administrative sub-divisions in which the Census schedules had been collected. Various grades of clerks were employed, the highest grade being allotted to the coding, which required a certain amount of technical knowledge and individual intelligence, namely, the coding of the replies to the occupation and industry questions. The other kinds of coding did not call for any considerable technical knowledge and the code numbers most frequently in use were easily committed to memory.

On the completion of the coding, the schedules were sent to the punch operators, who proceeded to transfer the coded information to the cards. The limitation of the size of the card, and the amount of the information collected at the Census, necessitated the use of three cards. The Personal Card (Fig. 2) was used for recording the particulars relating to individuals, one card being punched for each person enumerated. The Fertility Card (Fig. 3) was used for recording the particulars regarding married couples enumerated on the same schedule. The schedules were gone through twice, first for the Personal Card and a second time for the Fertility Card. The punching of the Population Card (Fig. 1) did not necessitate a third perusal of the schedules. This card was punched from the summaries of the schedules compiled by the local enumerators.

The punching operators were recruited from young girls just leaving the elementary schools. The full staff employed numbered 170, working in two shifts of six hours each. After a period of training this staff soon became efficient. The general average number of Personal Cards punched per hour showed a steady increase throughout the work, as will be seen from the following table :—

PERSONAL CARDS.
Week ending— Average per Hour.
   5th August, 1911 127
   2nd December, 1911 225
   6th April, 1912 327
   24th August, 1912 369
   21st September, 1912 417

Under the system of verification adopted, all the work of the punch operators was examined until a high standard of accuracy was attained. Two methods of verification were employed. At the outset, the cards were read by a punch operator to a female clerk who had the original records. At a later stage female clerks compared the cards with the records without the aid of the punch operators. As the operators became more accurate a greater portion of their cards was passed unchecked, except for periodical examination for the purpose of ascertaining whether the usual standard of accuracy had been maintained. A similar system was adopted when checking the Fertility Cards, but the punching staff employed at this stage of the work was quite expert, having had considerable experience in the punching of the Personal Cards. The standard of accuracy was found to be high and a large proportion of the work was not checked. As an incentive to good work, increases in the wages of punchers were given for speed and accuracy in punching.

As soon as a sufficient supply of the Personal Cards had been punched and checked, the sorting and counting of the cards were undertaken. The Sorting Machines were operated by boys who had just left school. After the cards were sorted and boxed away, all were proved by the clerk in charge of the Sorting Machine operators. As a further precaution the clerks in charge of the Counting and Tabulating Machines also checked the order of the cards at the time of counting. An experienced clerk was in charge of each Counting or Tabulating Machine. He was responsible for seeing that the correct cards were put into the machine in the right order, and for recording the results. A boy assistant was also allotted to each machine, and it was his duty to fetch the boxes of sorted cards from the filing racks, assist in proving the cards, feed and operate the machine, and return the cards to the racks. On the completion of a counting process for an administrative area, the result sheets were scrutinised and any apparent discrepancy investigated before the cards were re-sorted for the next counting process.

The total population of England and Wales, as ascertained at the Census of 1911, was 36,070,492. As previously stated, a Personal Card was punched for each person enumerated. The multiplicity of combinations into which the cards had to be sorted preparatory to the counting necessitated the same cards passing through the machines many times. From the following table it will he seen that the preparation of the tabulated results from the Personal Card was equivalent to the sorting of 344,278,000 cards, or an average of 95 passages of the 36,070,492 cards punched, and the counting of 147,950,000 cards or an average of 41 passages. Corresponding figures for the Fertility Cards show that the 6,136,605 cards punched were passed through the Sorting Machines 203 times, through the Counting Machines twice, and through the Tabulating Machines three times.

Total
Number of
Cards
Punched.
Card
Passages
through
Sorting
Machines.
Average
Number of
Passages.
Card
Passages
through
Counting
Machines.
Average
Number of
Passages.
Card
Passages
through
Tabulating
Machines.
Average
Number of
Passages.
Personal Card 36,070,492 344,278,000 9.5 147,950,000 4.1
Fertility Card 6,136,605 124,642,000 20.3 12,241,000 2.0 18,428,000 3.0

The figures in the above table are not materially affected by the work involved in sorting and counting the 52,000 Population Cards.

Population Hollerith Card (Figure 1)

POPULATION CARD (FIG. 1).


Personal Hollerith Card (Figure 2)

PERSONAL CARD (FIG. 2).


Fertility Hollerith Card (Figure 3)

FERTILITY CARD (FIG. 3).

Next Selection Previous Selection