Use of the Zenith Telescope in Determining Latitudes. 203 
From this, results the mean probable error of any number of 
observations upon a pair of stars, by dividing the value just found 
by the square root of the number of observations. 
We are now prepared to estimate the error of the places of the 
stars used ; for the discrepancies which they show are compose 
of errors of observation and of errors of the catalogue,* if the 
computations have been rightly made. form. in which 
nen is made is given in the annexed table, with the 
esults 
Number of Number of Star Mean of | Mean 
Pairs. in each pair.) latitude. ant (aw zy 
B. A. Catalogue. © | a hee 
6368 6427 37°92 | 34//-50 3°42 11°6964 
If 6497 6522 | 33-92 58 +3364 
il 6571 6583 36:20 1-70 2"8g00 
iV 6673 6712 41:74 724 52-4176 
V 1 'G6745 “1 G754 | “3325 25 50625 
VI°| 685r | 6928 | 28:42 6-08 36-9664 
VII 6983 34:58 123 0064 
Vil 7008 7076 33-27 08 151 
IX 7100 7171 34:55 05 “002 
XXXII 953 1043 S18 62 3846 
33. | 1138-66 | 2(a—2)* 169°6906 
Mean lat. = 43° 45! 34!-5o} 5M? 44418 
DHE Tt. 165-2488 
For x—1 pairs. 
3 x(a—x)?—=M? 165-2488 Log. =2-21814 
e Catalogue = 2-2) 5 = ne =1-50515 
0:71299 
‘4549 “ =9°65792 
e? Cat. =037091 
e Cat. =1/-53=0°18545 
Probable error of position of one pair of stars, 
given in B. A. Catalogue, . ‘ : = 1-63 
he four pair of stars hav- 
duce the average Oata- 
latitude but 
_ The omission of the observations on t 
ing the greatest probable error, would re 
b he omission of five pairs, 
able errors, would but slightly change the pro 
Catalogue, (to 0-93), while it would affect t 
+0”-32. The effect of a few very discordant wesults on the 
Probable error is not peculiar to these results, or confined to this 
Class of observations. : 
_A general comparison of the results obtained in the Coast Sur- 
vey,—as I propose, on some other occasion, to show,—confirms 
iccihis supposes he of - rometer value to be too small to enter into the 
