96
or had himself entered up the particulars. Further columns in the book gave the totals of each sex from whom particulars had been obtained in each dwelling. These columns provided a means of making a rapid survey of the population and were used in preparing the preliminary report.
During the month of March the schedules and Enumerators' Summary Books were subjected to a careful scrutiny, which in some cases involved reference back to the person making the return.
Unused and spoiled forms were discarded and the schedules made ready for indexing. The schedules were checked with the summary books and totals extracted. Summary Books were not employed in the New Territories nor in the enumeration of the Population Afloat so that in those cases the totals were taken direct from the schedules. A preliminary report was compiled and forwarded to Government on April 6, 1931, one month from the date of the Census. It was issued to the Press and published by them on April 17.
It is interesting to note how far this hurriedly prepared estimate was authen- ticated by the more deliberate totals arrived at later. The preliminary total for Kowloon was nearly correct, the error being less than .005%; the totals for the New Territories and the Population Afloat were also as correct as could be expected, as these were obtained by a rapid count over the schedules themselves, the errors being less than 1.69% and 1.55% respectively.
A serious error, however, crept into the Hong Kong total as regards the areas in the South of the Island. Though the totals of each page in the Summary Books were carried forward to the next page, the clerk responsible for these areas totalled the figures given at the bottom of each page. The numbers for the South of the Island were more than doubled by this error, having been included in the preliminary report as 28,900 whereas the true total was 11,155 only. There has been considerable development in these areas during recent years and the total, though it showed a large increase since 1921, did not at first appear absurd. There were a few compensating errors elsewhere so that the net excess in the preliminary estimate for the Island of Hong Kong amounted to 15,319 or 3.74% of the true total.
(b) Tabulation Procedure.
The rapid procedure adopted for the purpose of the Preliminary Report could not, however, be applied to the detailed contents of the census returns, each entry of which had to be separately examined and classified before any expression could be given to the information obtained.
The problem of the tabulation of the raw material obtained on census schedules and the orderly presentation of the results has always been one that demanded careful consideration. Prior to 1911 in Hong Kong the laborious method was used of obtaining all the information required by searching through the schedules them- selves for the answers to each question. In 1911 the slip or card system was first employed. The form and size of the cards or slips used must vary according to the extent of the information furnished by the schedules. Size and form being decided on, a separate card is allotted to each person enumerated, and on the card there are recorded all the particulars to be tabulated concerning that person. With the cards thus filled up, the work of tabulation is performed by simply sorting the cards under any given head and counting them. Generally, cards of different colours or shapes are used to indicate the more important divisions such as race and sex; the tabulation under these heads being thus carried out by simply counting, without read- ing. the cards of different colours or shapes.
In recent years mechanical tabulation for census statistics has been perfected and is used in many countries of the world. The Superintendent, while on leave in England in 1930, had an opportunity of seeing the machines in operation at Somerset House, London. As a means of transferring information from schedules