CHROMOSOME NUMBER AND PAIRS IN AMBYSTOMA 191 



Summary. The combined instrumental and personal errors 

 are reduced to practically zero by averaging several measurements 

 made upon different drawings. The errors due to twisting of 

 chromatids about one another are largely neutralized and are 

 considered indi\ddually later. Therefore, except for possible 

 unequal contraction and stretching of homologues, only the 

 measurements of those chromosomes which are foreshortened 

 contain appreciable errors. It is thought that these errors, 

 after corrections have been made, probably do not in any case 

 exceed 1 mm. The presence and amount of error due to unequal 

 contraction and stretching in any particular chromosome is an 

 uncertainty, but if existing would probably be greater in the 

 longer chromosomes. 



c. Results of measurements. 1. Criteria for determining pairs. 

 Before considering the results of the measurements, it seems 

 desirable to state what the criteria are that will demonstrate 

 whether pairs (p. 185) are present among the chromosomes at 

 these particular stages of mitosis. In the absence of definite 

 minute morphological characteristics, such as a repeated occur- 

 rence of marked granules, constant in position and size, which 

 Wenrich ('16) describes for certain Orthopteran chromosomes, 

 the next most exact criterion for determining the presence of 

 pairs in diploid cells would be a duplicate series of chromosomes 

 of equal volume. But in these chromosomes trustworthy volu- 

 metric determinations cannot be obtained, for the above- 

 mentioned intertwining of the chromatids and the stretching of 

 the chromosomes would cause variations in diameter which 

 could not be measured accurately, and these errors would be 

 cubed in the volume. Consequently linear measurements, sup- 

 ported by form, have been chosen as giving more trustworthy 

 data. 



Upon this basis, in order to constitute undeniable evidence 

 that the chromosomes form a duplex series, there are two con- 

 ditions which should be met. First, when the chromosome 

 lengths are plotted in a graph (e.g., figs. 33 to 37), they should 

 definitely associate themselves in twos of equal lengths. Second, 

 the differences in length between successive pairs, as indicated 



JOrRXAI, OF MORPHOI.OC.V, VOL, 33, XO. 1 



