472 DAVID H. DOLLEY 



taiice or indeed partially eliminates them. For two micra sections, 

 a fair approximation of duplication depends on judgment and 

 common sense; for one micron sections, it may be made much 

 more mechanical. For the latter, all polar sections were taken as 

 of full thickness and the few allowances that were made changed 

 the results only to the extent of a small fraction. By them- 

 selves, all that is accomphshed by the two micra series is to elimi- 

 nate the wider variations in the calculations of individual co- 

 efficient figures from the data of simple measurements used in 

 making the averages. By reference to the frequency-polygon 

 (text fig. A) it will be seen that the measurements of single 

 sections of cells taken as they came give coefficients ranging 

 from five to nineteen, though in table 2 the averages of these 

 same cells give fairly constant figures. The ten cells recon- 

 structed from two micra sections narrow the limits of variation 

 practically within one unit and confirm the accuracy of the 

 method of average measurements. The five cells reconstructed 

 from one micron sections in turn e&tablish finally the essential 

 accuracy of the grosser method of reconstruction and prove 

 beyond question that the ordinarily simpler method of average 

 m.easurements is sufficiently reliable. 



3. The uncertainty of uniformity in cutting the dendrite. 

 The general rule stated for measurements was followed, and 

 usually, with three dimensions to guide, the chance of discrepancy 

 is minimized. In the case of cells whose dendrite bifurcates 

 within a short distance, of which four examples were encountered, 

 it was found that unless the point of bifurcation was included, 

 the results were discrepant (figs. 2, 3 and 4). This confirmed 

 the practice of measuring such cells to include the crotch. 



4. The diflficulty of seeing the thinnest edge of the section (the 

 maximum cross section of the frustrum it constitutes) to trace it. 



5. The indefiniteness of the nuclear margin in the thinnest 

 sections, or, of more weight, the masking of the outline by peri- 

 nuclear chromatin, which in every resting Purkinje cell is almost 

 certain to be present in a small segment at least. 



6. Inequalities in thickness of microtome sections. This is 

 negligible because such series were avoided. 



