418 
I may mention that Capt. Parry in fixing his Long. at Port 
Bowen 88°. 54°, 55° E.—73°. 13‘. 39" N. Lat., made his Long. by 
6 occultations of fixed stars to differ from the mean of all his 
observations by 4° only. But by 
Moon’s transits (a large number) ... 2. 42". 
By eclipses of Jupiter’s satellites ...  2', 40%. 
By. chromometens<s.0sc--6.. acces 20”. 
By lumar distamees <.-ci.ccssseu. -ccteee 26". 
Again, he took 620 sets (310 E., 310 W.), and the extreme 
error is as much as 32. As 10 went to a set, it is possible that 
a part of the error arose from taking so many together. 
Lastly, I will mention one very accurate observation of the 
moon and Jupiter, May 10, 1875 (dist. 81°. 30°. 30"), which 
gave the Long. of Emmanuel College 45° E., or only 15° too much. 
Here the observed distance is about 10‘, or one diyision of the 
vernier too large, but the error is too small to notice. 
I may conclude by saying that in a few other cases of 
Lunars, when the moon has been nearly on the meridian, and 
at an altitude of at least 30°, I have not found any error except 
what might be properly ascribed to defects in my own power of 
observing, or a local error in the graduation of the circle; also 
that we may be sure that the large error in the first observations 
cannot well have been due to an error in the estimated amount 
of refraction, as the thermometer and barometer had been ob- 
served, Th. 48°, Bar. 29°. 9, about cancelling one another, (See 
on this point Ast. Soc, Month. Not. Vol. xxi. p. 58; Zach. 
Monat. Corresp. XXvu. p. 341; Shortrede’s Log. Tables (ed. 
1844), Introd. p. 12; Brinkley in Memoirs of R. I. A. xm. 
p. 170.) Nor has it escaped my notice that had the changes in 
the reduced distance (by Borda’s formula, I believe, see ante, 
p. 858) agreed with those in the observed distance, the error 
which I have discussed would not nearly have amounted to so 
much. 
