[ 539 1 
retardation of the plate in fwift motions, but from 
the hauling home of the log in flow ones. As for 
inflance; the log, to doits office accurately, ought to 
remain at red; in the water, whatever be the motion 
of the veffel. But even the keeping the line ftrait, 
and much more the fuffering the log to haul the line 
off the reel (as praffifed by many), will make the 
log, in fome meafure, follow the vefTel, and will be 
greater, in proportion as the time of continuance of 
this aft ion is greater ; and therefore the log will fol- 
low the fhip twice as far in going one knot, when the 
(hip is twice as long in running it. The confequence 
of this is, that a veffel always runs over a greater 
(pace than is (hewn by the log-line ; but that this 
error is greater, in proportion as the veffel moves 
(lower. It is this reafon, I fuppofe, that has induced 
the practical feamen to continue the diftance between 
their knots fhorter than they are dire died by the theory. 
Afterwards, in the fame fumrner, I made fuch 
another expedition, in a failing veffel, along with cap- 
tain Campbell of the Mary yacht, and Dr. Knight. 
Having prepared two of thefe machines as near alike 
as poffible, I determined to try, how far they were ca- 
pable of agreement, when expofed to the fame incon- 
veniencies, and ufed together. During the trial of 
thefe machines, one made 86,716 revolutions, and 
the other made 88,184. During this fpace, they 
were compared at ten feveral intervals. The revolu- 
tions between each interval differed from the propor- 
tion of thefe numbers, in the firft comparifon, one- 
nineteenth of the whole interval. The errors of each 
interval, in the other comparifons, were, in order, 
two-feventeenths, one-nineteenth, one-twentieth, one-y 
Y y y 2 fifty* 
