180 PROCEEDINGS OF THE ACADEMY OF [March, 



tic, 1.00 N. A. diaphragmed to .50 N. A. Stage Micrometer, Rogers 

 "B" Division, tenth two-thousandth inch, Zentmayer Filar Micro- 

 mater. Diameter of corpuscle No. 1, .00031770 inch. Variation from 

 first measurement, — .00000280 inch. 



These measurements were made with the utmost care, the mean of 

 ten readings of the filar micrometer being taken in each case, and 

 between the readings both the object and lines of filar micrometer 

 moved and the focus changed, so that each reading required independent 

 adjustment in all respects. The actual value of the micrometer 

 divisions was not determined until after all the readings on the cor- 

 puscles were made, to avoid any possibility of unconscious mental 

 bias. Fractions of an inch are employed because my best micrometers, 

 those with the sharpest defined lines, are so ruled. It is easy to cal- 

 culate the measurements to microns if desired, but this was useless 

 in present case where the measurements had no value save for com- 

 parison. 



It will be noted that the first two measurements, with illuminating 

 cones of 1.20 and .20 N. A. respectively, agree far within the limits 

 of probable instrumental error. In fact the small variation on cor- 

 puscle No. 2 is entirely a chance result, and I have experienced such 

 close results in but one or two other cases out of many hundreds of 

 measurements. The measurements on the second evening, under still 

 different conditions, strongly confirmed the previous results. The last 

 two measurements, each of but one corpuscle, were made principally 

 to test the accuracy of the stage micrometers used, and the greater 

 variation shown, which in the fourth measurement approaches what I 

 would consider the probable limit of instrumental error under the 

 conditions, might be ascribed to the use of different stage micrometers. 

 I doubt if this is entirely true, as all my micrometers have been so 

 thoroughly and repeatedly compared and studied, that any division 

 or divisions of any one of a half dozen or more might be used with little 

 chance of serious error, the true value of each, as recorded in my note- 

 book, being of course used in making the comparison, as well as the 

 same marked position on the division, as all stage micrometers vary 

 materially not only in different divisions but in different parts of same 

 division. The best have a horizontal line on which all comparisons 

 should be made. It is more likely that the comparatively large varia- 

 tion in the fourth measurement is mostly due to the employment of 

 the Bausch & Lomb instead of the Zentmayer Filar Micrometer. 

 While the former is very accurately made, it has ruled lines instead of 

 spider webs, and therefore does not admit of such accurate definition 



