946 ME. H. H. BRIXDLr.Y OX REmoDrcED [Dec. 13, 



So far as the tarsus of the Blattidae is concerned, reference to 

 tables B and C, giving the ratios for the several joints of the 5-jointed 

 and 4-jointed forms, seems to forbid such an explanation of the 

 condition of the latter. This is evidently divided up in a manner 

 peculiar to itself. A like conclusion follows a comparison of the 

 actual length of the joints of two tarsi of the same total length 

 and from the same pair of legs of one individual, when one is 

 4-jointed and the other 5-jointed. And with regard to other 

 cases, a consideration of the descriptions and figures of the authors 

 Avhose work I have quoted does not support the view that we can 

 explain numerical deficiency on the ground that any partiadar 

 joint of the normal appendage is absent in such and such an instance. 

 The structure of the reproduced appendage being what it is, seems 

 to render this kind of explanation meaningless, as Bateson (3) has 

 already pointed out in commenting on the reproduced tarsus of 

 Periplaneta. There is perhaps more to be said for the x'mw put 

 forward by some, that these abnormal reproduced structures contain 

 the representatives of one or more joints of the normal limb fused 

 together and that hence arises the numerical deficieucy. But such 

 an explanation demands that a certain joint of the reproduced limb 

 should be equivalent in length to the sum of two or more joints of 

 a normal limb of the same total length. But in the case of the tarsi 

 of Blattidae the measurements already quoted show that here at least 

 such an explanation is inadmissible, it is true that the uumber of 

 individual cases included in the tables was not large, but it may 

 be held to have been large enough to demonstrate that it would be 

 exceedingly exceptional for the sum of any two joints of the normal 

 tarsus to even approximate the length of a single joint of the 

 reproduced tarsus, for the measurements given contain no example 

 of this kind. We may suppose that Ji and J^ correspond with/j 

 and j. respectively on account of their structural characters and 

 position, but there is nothing to establish that J^ and J^ represent 

 either {j.,-\- j-^-\- j^ ov j.-^-\-{j^-^ j^ — a result which shakes confi- 

 dence in the identification of the longer proximal and terminal 

 joints with those of the normal. This matter also has been already 

 discussed by Bateson in the place cited, and it is enough to add 

 that his conclusion that the four joints of the reproduced tarsus 

 collectively represent the five joints of the normal, which was based 

 on measurements of Periplaneta only, is borne out by those of 

 StyJopijga made more recently. 



The view that such reproduced structures should be looked upon 

 as intrinsically on a different plan from the normal structures they 

 replace, rather than as abortive attempts at the exact reproduction 

 of those normal structures, finds support not onl}^ on the grounds 

 already set forth, but in some cases at least from the closeness of 

 the variation of their individual parts. In the case of Periplaneta 

 americana measurements of the lengths of the tarsal joints were 

 made in 115 normal and 115 reproduced tarsi. These tarsi were 

 all from the third pair of legs of adult individuals. The total 

 length of each tarsus was reduced to 1*000 and the lengths of the 



