Presidential Address. 
421 
time-piece is by means of the transit instrument. This instrument, as it 
were, forms the indicator, or "hand," of the clock, whose " face " is the 
sky itself, while the stars correspond with a series of irregularly placed 
graduation marks on the dial. 
With a transit instrument on stable foundations in conjunction with 
data furnished by the various nautical almanacs, it is not a difficult matter 
at any time when weather permits to obtain time determinations with a 
precision exceeding that quoted above as illustrative of the precision 
attained in the daily time service. 
Our usual practice for this purpose is to endeavour to secure an 
accuracy which would exclude errors of one-tenth part of a second, and 
doubtless this precision is attained in the majority of cases, except where 
long-continued cloudy weather prevents a direct reference to the sky and 
forces us back on the use of our clocks. 
When, however, a higher degree of precision in transit-observing is 
desired, difficulties of a complex character at once begin to present them- 
selves, and it is with the nature of some of these difficulties and the 
methods adopted for overcoming them that I wish to deal. 
In the first place, it is found that two different observers observing 
almost simultaneously with the same instrument under conditions as 
nearly similar as can be reproduced, persistently differ to a small but 
decidedly appreciable extent in their determinations of clock errors. 
Two methods of observing have been regularly used until recently. 
In the older method (known as the " eye and ear " method) the observer 
listens to the beats of his clock, counting off the seconds, and simul- 
taneously watches the image of a star travelling across the field of his 
telescope, and estimates to the nearest tenth part of a second the instant 
at which the star transits across a spider-web fixed in the focal plane of 
the telescope. In the more modern, or galvanic method, the beats of the 
clock are automatically recorded on a travelling tape or rotating drum, 
while the instant at which the star reaches the web is similarly recorded 
on the same tape (or drum) by the observer tapping a key. 
A comparison between the two records, from which the error of the 
clock may be deduced, can then be made at leisure by a method which 
admits of higher refinement, if necessary, in the subdivision of a second. 
Both methods are found to be liable to discordances of a similar 
character. It is not uncommon with either method to find two ex- 
perienced observers whose determinations persistently differ by as much 
as a quarter of a second, a quantity by no means to be disregarded when 
even a precision of only one-tenth part of a second is sought. 
A far higher accuracy than this must be aimed at in observations 
required, for instance, in the precise determination of terrestrial longitudes, 
pr in observations required for correcting the tabular positions of stars as 
