6 First Experiments. 



which provided against every possible source of error. 

 This could only be satisfactorily accomplished by 

 watching the development of the gall of each species, 

 from the time when the ^%'g was laid until the gall 

 reached maturity. Unfortunately, however, this peculiar 

 difficulty exists, that the most important phase of de- 

 velopment, viz. that during which the eggs of the fly are 

 buried in the bud or tissues of the oak, must unavoid- 

 ably be hidden from direct observation. Indirect 

 observation alone is available, for any actual examina- 

 tion of the eggs when laid must necessarily end in 

 their destruction. Thus if a gallfly lays its ^%% in 

 a bud, one can predict with certainty what gall will be 

 produced, so long as care is taken that the same bud is 

 not pricked either before or afterwards by another fly. 

 Breeding experiments must therefore be so arranged 

 as to enable each species to be isolated and watched 

 while actually depositing the ^^%. 



For this purpose I had a quantity of little oak trees 

 planted in numbered pots, each pot serving for experi- 

 ments with flies of the same species. The experiments 

 were made in a room, and the flies were carefully 

 watched from the time they were placed upon the tree 

 until they began to prick the buds : those buds which 

 had been undoubtedly pricked were then marked by 

 means of a thread tied around them. It was naturally 

 impossible to go on watching the flies for many hours 

 together, and I adopted the plan of covering the trees 

 over : in this way flies placed on the trees were pre- 

 vented from escaping, and at the same time those of 

 other species were kept away. At first I used glass 



