152 



MEMOIRS NATIONAL ACADEMY OF SCIENCES. 



Vol. XII, 



Indeed, the extremely generalized form of the larvae of this genus is clearly shown by the fact that in P. cecropia, 

 and all the other more specialized and hence later genera, there are only five tubercles on the eighth abdominal segment, 

 those corresponding to the two middle ones of Saturnia having, probably during embryonic growth, coalesced. The 

 embryos of these moths should therefore be examined shortly before hatching to ascertain whether this be not the 

 fact. Meanwhile it is not unreasonable to suppose that all the more specialized genera must have been derived from 

 a Saturnia-hke ancestral form, i. e., a larva of cylindrical shape, with all the tubercles, whether thoracic or abdominal, 

 of the same size, shape, and color on all the segments; those on the eighth abdominal segment being of the same number 

 (six) as on the segments in front. 



The single median tubercle on the eighth abdominal segment of the more specialized Saturnian larvae represents 

 the "caudal horn " of Sphinges, Bombyx mori, and the Notodontian genus Pheosia, and is evidently the result of fusion 

 before the end of embryonic life of what were originally two separate tubercles, like the two separate ones of Saturnia. 

 We are thus able to confirm the suggestion of W. Miiller, who first identified the "caudal horn" with the two dorsal 

 tubercles on the eighth abdominal segment of the Saturniidae. 1 



Thus as regards the tubercles the species of Saturnia axe on the same plane with the embryo, just before exclusion, 

 of the more highly specialized forms of the group Attacinae. The great size of the Attacinae, particularly Attacus atlas, 

 appears to be a sign of recent specialization, and the small size of Saturnia, aside from its other features, suggests that 

 it is a generalized form, not departing greatly from the normal size of the members of the superfamily Bombyces. 



And here an interesting problem in zoogeography occurs. Are the species of Saturnia (in the restricted sense) — 

 three in Europe, and two in the Southwest and Pacific coast of North America, occurring where the Attacinae do not 

 exist at all, or only rarely — the relics of a Saturnian fauna from which the groui^ Attacina; has been eliminated by 

 geological extinction, as the sequoia, cypress, magnolia, and other Tertiary plants have been rendered extinct in 



Europe, or may the view be taken that the Attacinae have never had a 

 foothold in Western Eurasia and North America? 



Should we use the characters drawn from the number and arrange- 

 ment of the tubercles of the larva in classifying the Saturniidae, we 

 might divide the family into two groups, as follows: 



A. Six tubercles on the eighth abdominal segment; the tubercles 

 in general over the body all of the same size. Generalized forms. 

 Subfamily 1. Saturniinx. 



B. Five tubercles on the eighth abdominal segment, the median 

 one double; the tubercles in general more or less differentiated or 

 specialized in size and color. Specialized forms. Subfamily 2. 

 Attacinx. 



An interesting series of parallelisms may be observed in com- 

 paring the early and later stages of the larvae of this family. For 

 example while the late embryos of the Attacinae are perhaps paral- 

 leled by the fully grown larva of Saturnia, the fully grown larva of the most or one of the most generfllzed Attacinx, 

 Samia, is on the same plane of specialization as the larva of Callosamia in its third stage. 



[The following appeared in Psyche, March, 1902, p. 321:] 



The subfamily Saturniinae, characterized by having six separate tubercles (the two median ones being separate) 

 on the eighth abdominal segment, comprise the following genera: Perisomena, Cricula, Saturnia (I can not see that 

 Calosaturnia mendocino differs from Saturnia), Heniocha, Loepa. 



The following genera belong to [Attacinae] * * * beginning as heretofore with the most generalized forms, 

 the exact sequence being subject to further modification : Copaxa, Opodiphlera, Tagoropsis, Syntherata, Rhodia, Rinaca, 

 Neoris, Caligula, Graellsia, Argema, Actias, Tropxa, Antherxa, Telea, Metosamia, Callosamia, Samia, Epiphora, Philo- 

 samia, Rothschildia, Coscinocera, Attacus. 



This subfamily is divided into several, at least five groups of genera in two series, for example, a Copaxa group, 

 an Antherxa group (Antherxa, Telea, Metosamia); a Samia group (Samia, Epiphora, Callosamia), a Tropxa group 

 (Graellsia, Argema, Actias, Tropxa), and an Attacus group (Rothschildia, Philosamia, Attacus). 



Whether Rhodia, Rinaca, and Neoris belong with Loepa, which has six tubercles on the eighth abdominal segment, 

 or with Copaxa which in stage I has but five, the median one being double, remains to be seen after we know more 

 of their larval forms. 



Fig. 16. — A, a dorsal view of the end of the body of 

 the larva of Saturnia carpini; B,the end of the body 

 of 5. pyri, copied from Duponchel. 



i W. Miiller, Siidamerikanischc Nymphalidenraupen, 1886, pp. 249, 250. Miiller remarks: 



" So erscheint os berechtigt, fur das Schwanzhora der Sphingidte die gleiche Genese anzunehmen wie fur den unpaaren Dora der Saturniada 

 auf 11. Beide sind entstanden aus den Stiitzgebilden dcr beiden Borsten 1 aui Segment 11. . . . Weiter finden sich bei einer Raupe, augen- 

 scheinlich den Saturniden angebbrig, in eincm friiheren Stadium Sds auf 2. 3, Ds 11; mit der niichsten Uautung verseh-winden die sammtlichen 

 Doraen. Bei Brahmen ledercri finden sieh im i. ( .) Stadium JOs 11, Sas 2-10. 12. Sst 4-11, von welchen Dorncn die Ds 11, Sds 2. 3 stark entwiekelt 

 dicanderen klein. unseheinbar sind. Im 4. (?) Stadium sind die Da 11, Sds 2,3 wohl entwiekelt. die anderen Doraen sind kaum nachweisbar. 

 Im letzten Stadium bieibt nur sine Warzc an Stelle des Be 11; es erhalt sich also der Rest von Ds 11. am langsten. 



"Mir schcinen alle dicsc Griindc zur Annahmc zu drangen, dass das Schwanghom der Sphingiden der Rest einer reicher entwickelten Bedor- 

 nungist, jiner Bedornung.nic vielleichl mit der heutigen der Saturniden auf gleichen Ursprung zuruckzufiihren ist. so dass das Schwan-hom 

 der Sphingiden und der Ds dorn der Saturniden im vollen Sinn homolog sind." 



See also E. B. Pouiton in Trans. Ent. Soc. London. 1SS5, p. 302, and in later papers; also A. S. Packard, Proc. Bost. Soc. Nat. Hist.. XXV 

 lS90,pp. )9. 103. foot-notes 1,2,3. Also compare our Figures 3-6, 8-10d', and the references to them in the text. Also Orote's N. A. Lepidoptera, 

 Bremen, 1880, pp. 16, 54. 



