360 G. H. KNIBBS. 
limits for any required ratio between the azimuthal and vertical 
components of motion, may be readily interpolated with sufficient 
precision for practical purposes. Very near the equator it will be 
desirable to observe the sun near its elongation rather than at a 
minimum altitude, if the altitude of elongation be not too great. 
It is of course never desirable to make observations for azimuth 
at great altitudes because level and collimation defects then enter 
into the result with large factors. 
There is a source of persistent error in meridian determinations 
by solar observations, the amount of which cannot be accurately 
defined, but might perhaps be fairly well ascertained by a sufficient 
comparison with the results of stellar observations. This error, & 
consequence of the impossibility of obtaining complete data for 
the evaluation of refraction, and of the imperfection of the refrac- 
tion theory, enters with a necessarily somewhat large factor into 
the results, and cannot be removed by combining morning and 
afternoon observations, because the physical conditions are asym- 
metrical. In the case of stellar observations the physical conditions 
for observations east and west of the meridian are frequently nearly 
symmetrical, and the final results consequently more reliable. 
The possibilities generally of meridian determination by solar 
obgervasions mae not, so eer as Lam aware, been fully investigated, 
an f this question is still a desideratum 
in F geodasy. Where extreme precision is not required, the solar 
methods are very convenient, and involve but little loss of time. 
How far the application of more rigorous methods of reduction 
will be justified, can be ascertained only by a careful criticism of 
results: the uncertainties of refraction, and imperfections of 
‘definition are so serious that it is still a problem whether the 
results can be materially enhanced in value. 
