CHROMOSOME NUMBER AND PAIRS IN AMBYSTOMA 187 
described above and have to do only with relative lengths. This 
fact eliminates at once a number of errors which would otherwise 
be very serious. 
The instrumental errors. The possible instrumental errors are 
a) failure to maintain a critical illumination, b) failure to maintain 
a constant wave length of illumination, and c) errors inherent in 
the planimeter and opisometer. 
It so happened that the strongest light was obtained a little 
above the point of critical illumination, and might therefore 
cause an error in measurement. However, several chromosomes 
were drawn a number of times under both conditions and no 
perceptible difference was observed. Any slight error overlooked 
would be equal in the homologous chromosomes and would not 
interfere with a relative measurement. 
Farmer and Digby (T4) state that errors can arise from the 
use of varying wave lengths of light. The same optical equip¬ 
ment and illumination were maintained in all of my operations 
so that relative values were unaffected. 
In making measurements with the planimeter the polar arm 
was held rigidly stationary in two grooved blocks so that the 
tracing point moved around in a circle having a diameter of 33 
cm. The sharp tracing point was kept upon the median line of 
the chromosome figure by moving the drawing around (without 
slipping) into line with the path of this tracing point, which was 
used as a pivot for orienting the drawing. A constant, repre¬ 
senting the value of each of the divisions of the vernier, was 
determined by measuring a series of known distances on a straight 
line. 
The accuracy with which this instrument was operated is 
indicated by the fact that the average difference between the 
extremes of measurements made upon each of several drawings 
is 0.3 mm., the standard deviation, computed from the combined 
measurements ,of several drawings, is 0.17 mm. The measure¬ 
ments, obtained more quickly with the opisometer, are slightly 
less accurate, the average of the above extremes and the standard 
deviations being 0.4 mm. and 0.26 mm., respectively. Finally, 
the average of the nine measurements made upon the three 
drawings of each chromosome reduces this instrumental error 
