,54 Capt. Kater’s experiments for determining the 
in defect 1,2 divisions of the micrometer, or ,00005 °f 
an inch. And as this is the portion of the scale employed 
in ascertaining the distance between the knife edges, this 
difference must ultimately be subtracted to obtain the dis- 
tance of the knife edges, in parts of the mean value of the 
scale.* 
From the high importance which attaches to General 
Roy’s scale, as having formed the basis of the Trigonome- 
trical Survey of the kingdom, I was particularly desirous 
of comparing it with that of Sir G. Shuckburgh, in order that 
I might be enabled to give the length of the pendulum in 
parts of that standard which constitutes the foundation of one 
of the most important scientific operations ever carried on in 
this country. Fortunately, this scale was purchased at the 
sale of General Roy’s effects by Mr. Browne, who readily 
confided it to my care. From the mean of a number of com- 
parisons, I found the distance from zero to 39,4 of General 
Roy’s scale, equal to 39,40144 of Sir G. Shuckburgh’s 
standard.-f 
* From an examination of this scale by the late Sir G. Shuckburgh, it appears 
that the greatest liability to error is ,00033 of an inch, or as corrected by Mr. 
TrouOhton, ,000165 of an inch, the chances against which are as 9 to 1. 
f The very great difference between this result and that stated by Sir George 
Shuckburgh, in the Philosophical Transactions for 1798, renders it necessary 
for me briefly to detail the manner in which the comparisons were made. The 
two scales were placed in contact, and remained thus for twenty four hours; after 
which, sixteen comparisons were taken in the course of the day ; but these were rejected 
in consequence of the temperature'having increased six degrees during the operation. 
When the scales had been together forty-eight hours, sixteen other comparisons were 
made during two succeeding days, the thermometer remaining steadily at 70®. The 
greatest difference between any one of these last and the mean result, did not amount 
to four divisions of the micrometer. The mean qf the first set of observations ex- 
