186 
CHARLES L. PARMENTER 
under which these measurements were made in order to judge 
their value correctly. 
1. Type of cells. Only cells were used in which every chro¬ 
mosome was perfectly clear and, except as noted (p. 189), lay 
exactly level in the equatorial plane throughout their entire 
length. Only three cells (figs. 1, 3, and 9) of this quality 
were available, and these were polar views of early metaphase 
stages in cells of the peritoneum and lung. The chromosomes 
of one other cell (fig. 10) approximated this condition and were 
also measured. The care with which these cells have been 
chosen may be judged from the fact that they were the only 
suitable cells in material from over one hundred larvae con¬ 
taining large numbers of division figures. In material with chro¬ 
mosomes so long and so numerous it is not surprising that so few 
cells were perfect enough for measurement. 
2. Method. In addition to choosing cells with chromosomes 
of the above character, three different camera-lucida sketches of 
each chromosome were made on different days with extreme care 
at a magnification of 2633 diameters. Each of these sketches 
was measured three or more times along the median line with an 
Ott compensating planimeter modified for this purpose, or with 
an opisometer. These nine determinations obtained for each 
chromosome were averaged to represent its length. This method 
is important because the extremes of these nine measurements in 
about one-fifth of the cases may differ 1 mm. from the average 
(and occasionally more). This demonstrates that one measure¬ 
ment upon a single drawing might give rise to an erroneous 
difference in the lengths of the homologues of some pairs rang¬ 
ing from 1 to 2 mm., the actual amount depending upon the 
respective errors in each homologue. Averages largely eliminate 
this error. 
3. Sources of error. The various sources of error may be classi¬ 
fied in three groups: 1) instrumental errors, 2) personal errors, 3) 
errors inherent in the condition of the material. 
In the first place, it should be emphasized that no attempt 
has been made to determine the actual length of any chromo¬ 
some. These measurements have all been made on the drawings 
