410 On the Method of determining the Time, Error, 
judging from my own expenienets I have no reason to doubt that 
he has over-rated the degree of exactness which may generally 
be attained by the method which he recommends. 
Some of his altitudes were taken by refiection from quicksilver, 
but he remarks that he has with equal success employed pure 
Iimpid oil for that purpose. I have made many experiments on 
this subject, and I find I can generally make my observations more 
satisfactorily from oil than either from quicksilver or any thing 
else that I have tried. From the slightest motion, the surface of 
quicksilver is so affected with tremors that it is quite impossible 
to use it with advantage. ‘The Trinity-House School is on the 
second floor of a substantially built house, yet I have there ob- 
served the surface of a quicksilver horizon sensibly agitated by 
the tumbling down of an empty ale-barrel in the adjoining vard. 
Indeed I have frequently taken altitudes with perfect satisfaction 
from the surface of oil, when, from slight and sometimes imper- 
ceptible causes, the surface of quicksilver has appeared so tre- 
mulous that no dependence could be placed on a contact ob- 
served in it. Water I have found liable to the same objections 
as quicksilver, though in a less degree. Treacle and other semi- 
fluids are subject to have their reflective power destroyed by dust, 
&c. and when their surface happeus to be disturbed, it requires a 
considerable time to settle. If the reflecting surfaces are unco- 
vered, as mine generally are, and at all exposed to the influence 
of the wind, oil is still more decidedly preferable. When taking 
the moon’s altitude during the day, I fiud it advantageous to pour 
the oil into a black receiver :—sometimes, as a convenient method 
of attaining the same object, I pour a little zvk into a white plate, 
and then pour the oil upen it. But when the altitude of a star 
is wanted, recourse must be had to a quicksilver horizon. 
The uniformity of General B’s results bears creditable testi- 
mony to the care with which he has made his observations ; but 
1 may be permitted to doubt whether he was justified in marking 
the time to tenths of asecond. The increments of time in the 
margin correspond, from his observations, to increments of 10’ of 
Forenoon. Afternoon. double altitude in succession ; and though no 
SECONDS. SECONLS. reasonable doubt can be entertained that the 
44-6 45 mean of the whole will be exact to a small. 
43 44 fraction of a. second, it appears sufficiently 
AA 45°3 obvious that the time of no single observa- 
45 43°6 tion can be depended upon to less than a se- 
44 43°6 cond. I am aware that a part of the trifling 
46 47 discrepances which will be observed among 
45 45°3 these intervals is to be attributed to the slow 
44 44 motion of the sun in altitude at the season 
44 495 in which the observations were made ; as it 
46 45°6 is 
Ee 
