METHODS 179 



therefore quite free from any doubts about relevance. In the next 

 paragraphs, I compare the advantages and disadvantages of these two 

 suggestions. 



The advantage of the single-handed method is that the biologist docs 

 not have to seek the co-operation of anyone. The data are all pub- 

 hshed, and later in this chapter, I give detailed directions about the 

 procedure of getting them. He can w^ork at his own speed, even if this 

 takes years. He would have to be very familiar with the animal or 

 plant he chooses to study, and he would have to know or to learn 

 statistical techniques. Unless he can borrow a calculating machine, 

 he would need the money from his own resources or from a research 

 grant to buy one, but, apart from this, the cost would be mainly 

 stationery, fares and postages on books. The main drawback is that 

 the patience to sit for hundreds of hours of computation and com- 

 pilation of data may not be common, but here I would point out that 

 the intricate operations have an interest of their own, and that real 

 flashes of the excitement of discovery will probably appear. Some of 

 the more mechanical operations can be carried out almost automatically : 

 I have often listened to a radio play while doing them. The idea that 

 any biologist likely to contemplate this work would be unable to 

 learn the technique is one that I have often heard, but cannot credit, 

 and I shall discuss this point later in this chapter. 



The advantage of the combined effort, electronic computor scheme 

 seems to me to lie mainly in the reduction of some of the drudger)^ 

 but also in the great economy resulting in the sharing of the work of 

 collection of the meteorological data among so many items in the list. 

 For example, all the events in May could, as a first approximation, be 

 supposed to be influenced by the same weather factors. Moreover, 

 some of the events would probably turn out to be strongly correlated 

 with each other, but some would not. This would suggest that those 

 that were correlated shared the same external influences, and this might 

 lead to the discovery of what these were. The drawbacks are : the 

 difficulties of getting co-operation between biologists interested in 

 many different things, and the high cost of computation. I was told 

 that for my little portion alone, although the actual computation 

 would take only five minutes, it would take an expert operator three 

 weeks to transfer the data to the operating tape, if I needed twenty 

 variables at twenty levels each. The cost would have been from ;£200 

 to ;(J300. With all the items in the phenological Hst to do, only one 



