148 
The Application of 
the corpuscles under a magnifying glass by means of a transparent 
scale ruled in hundredths of an inch on a thin strip of horn. The 
accuracy of this scale is of course determined by comparison with a 
standard. To avoid parallax, I tui-n the ruled side of the scale 
down so as to bring it in contact with the varnished film. It will 
very generally be found that the dried corpuscles are not perfectly 
round ; the longest and shortest diameters must then be measured, 
and the mean taken. As the image of each corpuscle is measured, 
a dot of water-colour is put on it, that it may not be measured a 
second time. This is readily washed off when the work is done. 
Extremely deformed corpuscles, and those which are obviously 
turned on edge, are not measured ; but no others are omitted. As 
each corpuscle is measured it is entered on a check-hst, which 
shows, when all on the negative have been measured, the number 
of corpuscles of each size. The sum of all the values, divided by 
the number of corpuscles measured, gives the average size of the 
images of the corpuscles in one one-hundredths of an inch. This 
average size, divided by the true magnifying power, gives the true 
average size of the corpuscles. To find the true magnifying power, 
I measure the distance of the lines of the micrometer from centre 
to centre on the negative, and divide by their true distance apart 
on the micrometer (that is, their nominal distance apart corrected 
by the ascertained constant error of the micrometer). The number 
of corpuscles on each negative, in my experiments, has ranged, 
with a single exception, from 50 to 175. Had the power always 
been just 1000 diameters, the measurements in one one-hundredths 
of an inch would have corresponded to one one-thousandths of an 
inch, and could be relied upon as correct to that figure. That is, 
writing the results in decimals of an inch, any error of observation 
would have been less than one significant figure in the fifth decimal 
place. With slightly higher magnifications, the results are still 
more accurate. When it comes to computing the mean of from 
50 to 175 such measurements, it can hardly be questioned that it 
is proper to carry out the mean results one decimal further, and it 
is not extravagant therefore to claim that the computed results 
are correct to the sixth significant figure. In the appended table, 
therefore, I give the average size of the corpuscles represented on 
each negative in millionths of an inch (together with the equivalent 
value in millionths of a millimeter, obtained by computation). 
I feel justified in claiming for the method above detailed, that 
it requires less time for an equal number of measurements, and that 
it is more accurate than any of the methods heretofore employed for 
the micrometry of blood-corpuscles. I also claim for it that it is 
capable of useful application in the micrometry of many other 
objects for which great accuracy is desirable. 
As to the time required, I suppose that twenty-five to fifty 
negatives, containing from 50 to 175 corpuscles each, can be ma(}e 
