20 MEMOIRS OF THE NATIONAL ACADE^SIY OF SCIENCES. 



In tlie final stage the horn shows some reduction, being shorter than in the previous stage. 

 There is no tendenc_y to fusion of tlie two dorsal warts on the ninth abdominal segment. 



The caudal horn in the Boni'iycidie sliows no resemblance oratSnity to tliat of the Ceratocam- 

 pidfe, and here I may state that the small family Bombycidw has no relationship with the Cerato- 

 carapidie or Saturniida?; on the contrary the family- is intimately related by its larval and pupal 

 characters with the Lasiocampidiv, and I am inclined to the view that the Boniliycida' have 

 actually originated from the Lasiocampidie, their larv;e having undergone a process of acceleration, 

 while the moths have 1)een modified by atrophy of the veins and mouth-parts. On comparing the 

 caudal horn of B<iiiil>y.f inor'i with that of Gdntrdjiai-Jui (Epicnaptera) Aiiieficana it was found to 

 lie similar in structure and armature, though that of Epicnaptera is Ijroad. short, flat, and conical. 

 On further comparison of the fully -grown larvae of the two genera I was surprised to find how 

 nearly allied they are. The head of BnnJiy.r mori (PI. XLIV, fig. 5) is of the same peculiar shape 

 as in the Lasiocampida% and densely clothed with long hairs; the}' are alike also in tiie nature 

 of tlie tergal region of the prothoracic segment, though B. mori is without the prominent lateral 

 tubercles so diagnostic of the larva' of Lasiocampida^ and Lyniantriida. In the ninth segment as 

 regards the shape of the suranal plate and of the anal legs, B. mori is closely similar to those of 

 Epicnaptera. I conclude, tiien, tiiat the Bombycida? being essentially lasiocampids in their 

 larval characters, as imagoes the group became modified by retrograde development and formed 

 a downward bent side-branch of the lasiocampid stem, ^^'e have here a clear examjjle of the 

 evolution by atrophy of one family fi-oni another. Dyar" states that the warts of B. mori iU'e 

 •• true warts of the typical lasiot'am})id pattern." He pla( es the family near the Lymantriidte. 

 To Professor Sasaki'' we are indebted for an account, with figures, of the first stage of the wild 

 silkworm of Japan, which Sasaki identifies as Theophiln imnularina, and its descendant or deriva- 

 tive Bomlitj.r tiiori. In both of thes^ forms there is already in stage I a single median wart 

 on the eighth abdominal segment. This shows that the larva has undergone, just as in Epic- 

 naptera, before hatching an acceleration of development as regards this pair of warts. After the 

 first molt the caudal horn is developed. Sasaki oljserves: '"After the second stage there are no 

 marked changes in both color and markings till the larva becomes mature." 



I have always supposed that the Bombycida? were more nearly allied to tlie Saturniida?, since 

 they have but three branches to the median vein and atrophied month-parts, but we now see that 

 these reduction characters are not diagnostic of any particular grouj) of families, but maj- occur 

 in any group as the result of disuse and loss of power of flight in the imago. Here, as elsewhere 

 in the insect and animal kingdom, the larval or postembryonic character as a rule, though there 

 are exceptions, aflord the truest guide to the phylogeny of a group. 



It is interesting to observe how the armature of certain species of Bombyx and allied genera 

 repeats in general appearance that of the ceratocampid genus Adelocephala, etc., though the 

 resemblances are surely cases of convergence, due perhaps to their living on trees whose twigs 

 are either tuljcrculated or spiny. 



The two most striking cases of mimetic analogy, parallelism, or convergence are seen in the 

 larva of Thenphila /(/f/Zo;*/, described and figured l)y T. Hutton. In the newly hatched cater- 

 pillar there is, he says, '* a small anal tubercle on the penultimate segment: thus far there is 

 scarcely a ditterence between it and tlie young Chinese worm."" This is evidently the same as 

 tlie incipient caudal horn of the larva of the same stage of Bomby.r inandarimi and B. i),iiri. 

 After the first molt the warts l)ecome converted into short, conical, fleshy tubercles or spines. In 

 the mature larva "there are two dorsal rows of long, l)lack, slender, and sharp-pointed spines, 

 connnencing with the fifth [second abdominal] segment. While the median spine is thus fused 

 or doul)le, the warts on the ninth segment of stage I become long spines aliout as large as the 

 caudal horn. 



"Oil the larvc-e of the higher Bombyces. Proo. Bost. Soc. Nat. Hist., XXVII, p. 140, 1896. 

 ''On the affinity of our wild and domestic silkworms, .\nnotatioues Zoologieae Japonenses, ii, Part II, Joiie, 

 1898. pp. 33-10. 



