14 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. I3I 



closely related to the cockroaches. An example is shown in figure 6 : 

 4, 5. There is some indication also, in various other primitive groups, 

 that there existed muscles of the type lils-epsz; if so, these too may 

 have contributed to the arrangements now seen in blattids, for they 

 could conceivably account for the lateral portions of the cruciate 

 muscles that run in some species from epSs to the transverse ligament 

 isps-iils or to the septum that has replaced it. 



b. Cruciate coxal muscles. — Mention of the cruciate coxal muscle, 

 epSz-cxiX (p), has been made in the preceding section. This muscle 

 is inserted together with the spinal posterior rotator isps-cxi (7), 

 from which it is morphologically distinct. However, species such as 

 P. australasiae, in which the origin of "epSg"-cxiX is far mesad, could 

 properly be described as having two definitive spinacoxal posterior 

 rotators, as was done by Maki (1938, fig. 6, Nos. 24, 25). In most 

 instances these two muscles may still be distinguished by the fact 

 that the true spinal muscle originates along the side of the spina ven- 

 tral to the other spinal musculature, whereas the muscle equivalent 

 to epSg-cxiX has a more dorsal origin, anterior to the transverse 

 muscle isps-epSz. Yet the distinction is not always clear ; and in some 

 blattids one or the other of these two muscles may even have been 

 lost. 



Serial homologs of the muscle epSg-cxxX do not occur in cock- 

 roaches so far as is known, but homologs with origins on the Us are 

 found in all three thoracic segments of larval Dytisciis (Speyer, 

 1922) and in larval Cybister. In larval Corydalus, which lacks such 

 muscles in the first intersegment, cruciate posterior rotators of the 

 second and third coxae originate on the corresponding furcal arms. 

 This shift in origin is easily understood from the fact that the furcal 

 arms are here fused with the succeeding Us, evidently, as judged by 

 conditions still found in some other Megaloptera such as Sialis spp., 

 in consequence of sclerotization along the line of the postcoxal liga- 

 ments 2ils-ju2, pis- fits. One infers from the position of the cruciate 

 coxal muscles in these primitive forms that the attachments of the 

 cruciate muscles of cockroaches on lils or epsg are more likely the 

 primary ones than any of the other variants observed in blattids. If 

 so, Cryptocercus and, for some strange reason, P. amcricana but not 

 its congeners have more nearly preserved the original condition. 



Cruciate promoters of the first coxa have been described from a 

 number of orders, and are apparently present in a much reduced state 

 in all cockroaches, although on account of their delicacy they have 

 escaped the notice of myologists. In the adult insect, which is the 

 stage usually chosen for dissection, they are extremely slender and 



