564 MR JOHN ALLAN BROUN ON THE HORIZONTAL FORCE 
changes after February 1843 have been on the average about 6-0, and that alter- 
nately plus and minus, the error may have been due to turning the telescope on 
its Ys through 90° or 180°, the error of collimation, being doubled in the latter 
case, and for a different wire in the former.* 
145. 1845, July.—Three of the weeks in this month seem to be affected by 
some error, similar to that noticed above; thus, taking the mean of each five days 
out of six (neglecting the first day in each week as not strictly comparable at 
different places), we have the following Table :— 
TABLE XXXVIII.—CuHaAnGES OF FIVE DAILY BIFILAR MEANS, 

k = 0:00014. 
Period. Trevandrum. | Singapore. | Hobarton. ise 
Sr div. Se. div. Bandi, Se. div. 
June 23-27. Mean 16°60 27°44 42°85 27:44. 
(June 80—-July 4) minus (June 23-27) — 2:82 0:53 2:64 2:70 
(July 7- 11) ,, (June 30-July 4) | + 1:08 4:31 0:55 218 
@uly. 142 18) 0, @uly vo mine | 8G 2:08 2-07 9-29 
(Guly 215025) 4, daw l49 Gals | — 408 1:03 4:43 4-04 
(July 28-Aug. 1) , (July21- 25) | +4. 0-24 081 2-21 0-81 
(Aug. 4- —8)_—«g,_ (July 28-Aug. 1) | — 0-02 4-48 2-03 1-47 
(Angyit- 15), (Awe, 4" 98), | 2-70 2-97 338 2-97 

The changes at Trevandrum and Hobarton agree very nearly, with the exception . 
of the mean for July 28 to August 1, which is two divisions greater at Hobarton 
than Trevandrum. A careful examination of the daily means at the three places 
induces me to conclude, that the error with the telescope occurred July 2; that 
the telescope was returned to its original position July 13; again moved July 22, 
and finally returned July 31. Ifwe take the error due to this movement (say 
turning the telescope round on its Ys) as before, at 6:0 or 4:3 scale divisions 
(1 se. div.=-00014), and correct the days on which the change occurred by only 
half this amount, we shall obtain the differences as in the last column of Table 
XXXVIII., which agree very well with those for the other places. 
146. In making this alteration it should be remarked, that errors at Singapore 
at the times noted are very probable (founding on many cases where I have been 
able to discover the causes of errors by this means); but no conclusions are based 
upon this correction, and it is not wished that any greater weight should be placed — 
upon the agreement of the means, corrected by this hypothesis, with those at 
* Errors from this cause were found to have occurred at Trevandrum. 

