18 MEMOIRS OF THE NATIONAL ACADExMY OF SCIENCES. 



Ill Citheronta regalis^ stage I (PI. LIV), the relations are much as in Eacles. The eaudal spine 

 is a nearly exact repetition of the twelve thoracic ones, endiiii;- in a similar bulbous body formed 

 by the fusion at their bases of the two terminal tubercles, each division or fork endino' in a thick 

 blunt rod. 



In stage II the caudal horn again repeats the now somewhat modified shape of the horns on 

 the thoracic segments (tig. 1), ending in two iuie([ual si)ines. 



In stage III the caudal horn is of the same proportionate shape and height (tig. 3). 



In the last stage the caudal horn is high and prominent, being as large as the subdorsal 

 thoracic horn, spinose, and distinctly forked at the end (rig. ti). 



In C. i<2)lendfj)is (PI. LV, rig. In. Ih /') the hoi'n is nuich shorter, l)ut still birid at the end; 

 in O. .sej/nJcrali's (rig. 4;^ -ih) it is much longer, but still ends in two minute tubercles; in ( '. Uincoon 

 (fig. 2, //) it is rather shorter and slendei'er than in the othei' species mentioned. 



Anisota differs remarkably from the other genera of its group in having no caudal horn on 

 the eighth abdominal segment, the primary setiferous tubercles / not l)eing crowded and fused 

 into one aftei' the rirst molt, the characteristics of stage 1 l)eing retained throughout laival life, 

 the four primary dorsal setiferous tubercles being arranged in a trapezoid. It .should be noticed 

 that the eighth al)dominal segment is not any narrower than the seventh, while the ninth is much 

 narrower, the body rather suddenly narrowing to the end of tiie suranal plate. Whether this 

 width of the eighth segment has caused crowding and the fusion of the tubercles / is a matter 

 worth considering, though in Adelocephala and V. ri'ijalis the eighth segment appears to be as 

 wide as the seventh. Yet it remains a fact that in Anisota there is no fusion of tubercles /, no 

 median dorsal horn, whereas there is a fusion on the succeeding segment and a resulting median 

 tuliercle. Its history is given on page 13 and the reader is referred to Pis. L-LIII, which 

 illustrate the development of this ninth uromeral caudal or median horn in Anisota. 



Here the question arises whether the lack of fusion of tubercles / on the eighth uromere is 

 due to inheritance from some notodontian ancestor, to reversion, or is simply the result of 

 disuse, resulting in reduction and partial atrophy. When we take into account the close 

 relationship between the moths of Anisota and Adelocephala, the incongruity between the larval 

 stages in respect to the armature of the end of the body seems remarkable. 



It is to be ol)served that the ninth uromeral tubercle or median horn in the final stage shows 

 its double origin in being bifid at the end like that on the caudal horn of the eighth segment in 

 other genera than Anisota. 



This ninth uromeral tubercle is well developed in all stages in I^acles and Citheronia 

 PI. LIV (fig. 6), but exists in a nuich reduced condition in Adelocephala and Syssphinx of the 

 first to the final stage. 



The caudal horn in the Notodontidiv. — In only two groups or families is the caudal horn a 

 nearly persistent characteristic conmion to nearly every genus of these extensive groups — I 

 refer to the Ceratocampida> and Sphingidsv — while the more specialized Hemileucidie and Satur- 

 niida? have a similar double median tubercle, which retains its rounded or clavate form. Else- 

 where the fusion of the tubercles / on the eighth uromere is a sport, aberration, or, better, a 

 mutation, which has become fixed by heredity. Such is the caudal horn of Bomhi/.r iimri, of 

 Endromis versicolora^ the earlier stages of Brahmaeida?, and of certain Notodontidie. 



Its appearance in this group we will first discuss, since their stages of evolution have been 

 examined and illustrated in my monograph of this family. In this group the larvae before the 

 first molt have the primitive arrangement of the four dorsal setifcM-rnis tubercles (/, //) arranged 

 in a trapezoid, as seen in the freshly hatched larva of Lophodonta, Pheosia. Dasylophia, 

 Hyparpax, Schizura, and Heterocampa. A caudal horn or tidiercle arises in the ontogeny of 

 the larva in four genera, i. e., Pheosia, Dasylophia, Hyjxirpax. and Schizura. 



The tubercle is most characteristic in Pheosia, being like that of Sphinges, Bomhy.vmori, and 

 ceratocampids, except that it is smooth and not solid and tuberculated. If the reader will turn 

 to PI. XX of my monograph of the Notodontida- and consult the description of the early stages 

 he will obtain an idea of the mode of origin of the so-called horn. In stage I the arrangement 

 of the dorsal and subdorsal setiferous tubercles is plainly drawn. The pair of tubercles / on the 

 eighth abdominal segment are, at the birth of the larva, crowded together, their bases forming a 



