METHODS 187 



have been for mc a practical impossibility I obtained permission to 

 borrow the volumes in sets as I needed them. All this mass of data, 

 both for the spawn dates and for the other variables, was transferred 

 to punched cards. In the case of the meteorological data I wrote 

 hardly anytliing, but found the necessary item on the page, and clipped 

 the cards according to the code. Various codes were used according 

 to the circumstances. I adopted i, 2, 4, 8, as the numbers of the holes 

 needed in combination to add up to 15, in preference to the more usual 

 I, 2, 4, 7, because although this code never needs more than three 

 clippings to register the number, they only add up to 14. To give an 

 example, I used Fahrenheit temperatures because the Centigrade scale 

 is too coarse a grouping, and to use a decimal woidd have taken 

 another hole. I ignored decimals of a degree, and subtracted 30 from 

 each reading. A temperature of 37-4 thus became 7 on the cards, and 

 was registered by punching holes i, 2, and 4. All these adjustments 

 can be done mentally, at the moment of punching. There are some 

 temperatures under 30 and over 45. To have dealt with these relatively 

 few readings by the code would have needed more holes, so I punched 

 all the holes, which acted as a warning to look for the actual number, 

 which I wrote on the cards. It is unwise to be too slavish in tliis matter 

 of punching — it may be simpler to write some of the data. For example, 

 I needed the mean spawn date on many occasions. To use the holes to 

 fnid the date on each card would have taken too long, so I wrote the 

 date as well as punching it. 



Having prepared the 2,734 cards, the next step was to set up the 

 correlation tables. This was done by first taking the cards, and sorting 

 according to one variable by needle. This is a well-known procedure, 

 and consists in passing the needle through the "8" hole of a bundle of 

 cards, as large as can be conveniently handled, say 100 cards. The 

 cards with this hole cUpped fall, the others remain on the needle. Each 

 of these piles is then sorted by putting the needle through hole "4, 

 and so on until all the cards have been sorted into piles, each corre- 

 sponding to one temperature. These piles were then secured by a tie 

 through a large hole near one corner of the cards, and put into a tray. 

 Each bundle was then taken and sorted in a similar way according to 

 the second variable. The cards in these much smaller bundles were 

 then counted and the number entered in the appropriate space in the 

 table. The small bundles were combined and replaced in the tray, for 

 the next table will require one of the variables to be the same, if the 



