THE GENETIC NATURE OF TAXONOMIC DIFFERENCES 267 



aedchfghii and aehgfbcdi, it is clear that the second might be derived 

 from the first by inverting the section bcde, and the third might then 

 be derived from the second by inverting dcbfgh ; but on the other hand, 

 there is no simple way in which the third arrangement can be derived 

 directly from the first. It is very interesting to find that the arrangement 

 which one is led to suggest as ancestral in the pseudoobscura stock 

 presents some similarities with that of the distinct species D. ?niranda. 



In the saUvary glands of hybrids between Drosophila species, large 

 sections of the chromosomes usually fail to pair at all; presumably they 

 have become too radically changed during evolution. An indication of 

 the kinds of changes occurring is given by the comparison of D. pseudo- 

 obscura and D. miranda,^ where pairing occurs between short homo- 

 logous sections which have been extensively reshuffled by inversion 

 and translocation; it is worth calling attention to the gross similarity 

 between the chromosomes in these two species, which gives no indica- 

 tion of the very considerable differences which really exist. Arguments 

 of species relationship must be accepted with great caution if they are 

 based only on similarity of metaphase groups. (Fig. 119.) 



Detailed investigation of the saUvary gland chromosomes within a 

 single Drosophila species often reveals the presence of very small 

 duplicated sections.^ The dupHcated section often hes quite near, or 

 even in contaa with, the normal section and is sometimes inverted 

 with respect to it, so that we get sequences of bands like cdefghfeijk or 

 cdeffeghijky etc. In the paired chromosomes these small sections may 

 be in contact with their homologues so that the chromosomes are bent 

 round into a sort of knot. Other minute changes would not be so easy 

 to discover, since, for instance, if an inversion is very small there will 

 not be room for the characteristic loop pairing to occur. But the presence 

 of the small dupUcations makes it very likely that small inversions are 

 also formed, and they may explain why in many regions the chromo- 

 somes fail to pair for no apparent reason in hybrids between species 

 such as D. melanogaster and simulans. 



Another type of chromosome change, which, since it leaves the total 

 balance of the chromosome complement unchanged, may be expected 

 to be involved in species formation, is mutual translocation, by which 

 chromosomes abcd,efgh become converted into abgh,efcd. In hybrids 

 between species differing in this way, multiple associations of chromo- 

 somes wiU form, giving chains or rings. For instance, in the above 

 case we can get an association as a chain abcd,dcfe,efgh,hgba and the 



1 Dobzhansky and Tan 1936. ^ Bridges 1935. 



