O. A. Merritt Hawkes . 139 



usually from the mid-dorsal region of the second abdominal segment. 

 The mid-dorsal spots had usually entirely disappeared before the upper 

 lateral series were affected, although there might be a simultaneous 

 diminution in the size of the spots all over the body. This suggested 

 that one was not dealing with spottedness (S) as a simple character but 

 as a compound of a large number of characters, each of which was related 

 primarily to a definite segment and subsequently to a segment as a part 

 of a longitudinal series. 



The above statement, " every variety of mating," must be qualified— 

 the RS condition represents the loss of anything between 1 and 140 

 (approx.) spots. On account of the small space and limited amount of 

 food at my disposal, I have had to experiment as if the S condition acted 

 as a heritable unit ; ideally, the matings should have been made between 

 each different type of reduction. But as the S and P types occur 

 naturally, the method of experiment used may be considered as a useful 

 and justifiable introduction to further detailed and more analytical ex- 

 periments. Also, as the lower lateral row of spots occurs in riciiii and 

 the upper rows are absent, may it not be that there is a connecting link 

 between the gens for the upper rows of spots, which has, somewhere 

 in the past, made them act as a unit, in appearance or disappearance, 

 independent of the lowest rows of spots ? 



Table I shows that the proportions among the larvae and imagines 

 are approximately the same, in spite of the numbers, which appear so 

 small when compared with those of the Japanese workers. This means 

 that, under domesticated conditions, there is no selective death rate as 

 regards spotted and plain larvae in the pupae : this in an important point 

 in deducing results from an animal which appears in three stages in its 

 life history and where, in artificial conditions, the death rate is so great. 

 It should not be forgotten however that, even in its native haunts, the 

 death rate must be very high. Thus Crampton(2) states in reference 

 to Philosamia cyrdhia (advena) — " the perfect imagines constituted only 

 166°/o of the whole number of individuals which entered the cocoons, 

 from which we may gain an idea of the severity of the conditions under 

 which the quiescent pupa exists." 



In the early stages the majority of deaths are due to the strain at 

 ecdysis, but in the later stages, deaths are mostly due to disease. An 

 attempt was made to find if there was a differential death rate for the 

 S, RS and P types during the first instars, but it was unsuccessful. 

 The larvae are all spotted when they emerge and become RS and P at 

 later ecdyses ; but whereas these changes take place at regular and 



