CHROMOSOME STUDIES 271 



present paper and in Studies III a class of variations that seem 

 to be of an abnormal type, which, as I have pointed out in Study 

 III, are also probably of a permanent nature. One of these 

 (figs. 136-147) is probably to be placed in the class of 'supernu- 

 merary chromosomes.' Up to this time the supernumerary 

 chromosomes described have been either extra sex chromosomes 

 or chromosomes in some way partaking of the properties of sex 

 chromosomes in that they are usually condensed. Here we prob- 

 ably have in the l^-valent portion of the large abnormal chromo- 

 some 1 the remains of supernumerary no. I's strung out in a 

 row (compound-supernumerary). Just what relation to varia- 

 tion they may have, it is difficult to imagine. A more impor- 

 tant representative of this abnormal class for matters of varia- 

 tion, it seems, is the deficient no. 4, whose measurement (Studies 

 III) shows it clearly to be about four-fifths the length of its mate, 

 which has the size normal for the halves of the pair in the species. 

 In such a chromosome we have the basis for the dropping of 

 unit factors from the germ plasm. This, I think, might explain 

 those characters which are evidently due to the lack of some- 

 thing, such as pigment or its producer in albino animals, and 

 come under the class of 'loss' unit characters. But such sorts 

 of variations (mutation) are not, it would appear, to be classed 

 with variations which mark off species from each other, such for 

 instance as distinguish Acridium granulatus from A. obscurus. 

 They are, instead, to be classed as minor variations occurring 

 within a species and possibly of little use in the production of 

 new species. It is not possible to consider them normal, for it 

 might be a question whether a race having such could be per- 

 manent until at least some additional compensating variation 

 occurred to supply what is lacking. Possibly there may be 

 some connection in this respect between compound-supernumer- 

 ary autosomes (1) of one pair and the deficient autosomes of an- 

 other pair. 



Our conclusion is that in the chromosomes of the germ cells 

 variations have occurred, as shown by taxonomic relations, and 

 may again occur, as shown by deficient and compound-super- 



