APPENDIX 5
Mechanical Method

In principle the method employed mechanically to convert quantities of foodstuffs into terms of nutrients was very similar to the manual method. The Hollerith punched card system was used.

Foods were classified into 68 groups.

A set of ‘ready-reckoner’ cards was prepared showing for each foodstuff the nutrient value of a unit (units were in ozs. 1/2 ozs., or table and teaspoons).

Cards were punched for each child indicating each foodstuff eaten during one day. The child had, on an average, 15 different foods per day, thus 105 cards were punched for each of the 426 children - a total of about 45,000 cards. These detail cards were hand-punched with the day-code, the food-code and the quantity. The individual child’s number and sociological data were gang-punched on to each set of 105 cards.

The 45,000 detail cards were mechanically sorted with the ready-reckoner cards on foodstuff and the quantity within each foodstuff group. Nutrient values on the ready-reckoner cards were then mechanically reproduced on to the appropriate detail cards.

The next operation was to sort the detail cards on a sorting machine to each child’s number, after which two tabulations were prepared which gave (a) a list of total quantities of every foodstuff eaten by each child during the week of seven days, and (b) the total daily nutrient intake of each child.

Comptometer totals from the forms were checked against tabulation (a). Tabulation (b) consisted of 2,982 lines (426 children x 7 days) and a summary card was hand-punched for every line of this tabulation. Thus these summary cards contained the sociological information relating to each child in the sample and his/her total daily nutrient intake.

Finally, the summary cards were sorted and tabulated to give the total nutrient intake of all the children within certain groups.

There was also a table prepared which showed for each of the 426 children the weekly menu and the quantity of the various foods that comprised the menu. For this purpose the original 68 foodstuff-groups were reduced to about 25. One card was hand-punched for each child with the menu and sociological details and these were analysed in the same way as the summary cards containing nutrient intake, except that their averages were computed per week instead of per day.

To sum up, the Hollerith machines used were:-

Hand Punches

Gang Punch

Reproducing Punch

Sorting machine

Rolling Total Tabulator

Some 48,000 cards were hand-punched, a further 95,000 either gang-punched or reproduced, there were about 400,000 card-passages through the tabulator. Use of a tabulator of lesser capacity - e.g. Hollerith E6/6 would have entailed about three times as many card passages as the rolling total tabulator.

To convert the foods into terms of nutrients rather more than half a million extensions were mechanically calculated in addition to very numerous totallings. Provision was made for ample internal checks to ensure accuracy as well as the comptometer totals made directly from the forms. This occupied three or four comptometer operators for a week.

The analysis was started in the middle of the October 1943, and all the results were available before the end of January 1944.