142 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 1 37 



the elements were still definitely muscular. Thus, the thysanuran pat- 

 tern may be ancestral to that found in certain pterygote groups, but 

 the present Thysanura cannot stand in the direct line of pterygote 

 descent. 



3. The Megaloptera are the only existing order with a pattern that 

 might be ancestral to that of the Isoptera. Inasmuch as there are good 

 grounds for postulating an entirely different origin for the termites, 

 the similarities observed here must be put down to the fortuitous reten- 

 tion by both groups of a number of the same primitive characteristics. 



4. The remaining orders have patterns derivable by simple reduc- 

 tion from both the thysanuran type and that of one or more other 

 orders, as follows : Plecoptera, from Blattariae ; Mecoptera, from 

 Odonata ; Phasmatodea, from Mecoptera and Mallophaga ; Hymen- 

 optera and Embiodea, from Orthoptera and Psocoptera ; Trichoptera, 

 from Blattariae, Megaloptera, Coleoptera, Psocoptera, Isoptera, and 

 Plecoptera ; Lepidoptera, from the same orders as Trichoptera, plus 

 Orthoptera and Trichoptera ; Ephemeroptera, from Orthoptera, Pso- 

 coptera, Neuroptera, Mecoptera, and Phasmatodea ; Thysanoptera, 

 from the same orders as Lepidoptera, plus Neuroptera and Lepidop- 

 tera ; Homoptera, from Thysanura, Blattariae, Megaloptera and Isop- 

 tera. To these should be added Coleoptera, Psocoptera, Plecoptera, 

 and Trichoptera, if the muscle described by Larsen (1945c) is isps- 

 lils; but only Orthoptera and Grylloblattodea if this muscle is isps- 

 eps. Hemiptera could be derived from all the foregoing orders except 

 Mantodea, Neuroptera, Ephemeroptera, and Thysanoptera. 



This tabulation reveals incidentally that, in general, the number of 

 possible paths of derivation increases as the number of spinasternal 

 muscles decreases, without necessarily indicating genetic relationships. 

 The need for other criteria therefore grows as the number of spina- 

 sternal muscles diminishes. 



5. The brachycerous Diptera, for which several species have been 

 studied, apparently lack all spinasternal muscles. 



The conclusions drawn above can be set down with confidence from 

 simple comparison of the spinasternal muscles of the different orders. 

 In proceeding beyond this point in the attempt to understand insect 

 relationships, it is necessary to resort to logical though hypothetical 

 reconstruction of presumed ancestral spinasternal patterns. A few 

 examples will clarify what can be learned in this way. 



Of the 14 muscles that constitute the spinasternal pattern of Pso- 

 coptera, only one, isps-iils, is missing from Orthoptera. However, 

 this same muscle occurs in 10 other pterygote orders and in the Thy- 



