SURVEr OF THE BRITISH ISLES FOR THE EPOCH JANUARY 1, 1891. 547 
Table XXTII. 
IS86. - 
1891. 
Difference. 
Means. 
A 
Lairg. 
- 68 
- 95 
+ 27 
Golspie. 
- 99 
— 66 
- 33 
+ 291 
Thurso. 
- 88 
— 125 
3/ 
± 45/ 
Wick. 
-1-182 
+ 97 
+ 85 
B 
Kyle Akin. 
-142 
-105 
- 37 
- 291 
±109/ 
Gairloch. 
-■S89 
-510 
+ 121 
Soa. 
-365 
-105 
-170 
C 
Mablethorpe. 
-h 96 
+ 208 
-112 
+ 71 
±81/ 
Lincoln . 
+ 124 
+ 31 
+ 93 
Gainsborough . . , . 
+ 87 
+ 48 
+ 39 
D 
Oxford. 
+ 102 
— 25 
+ 127 
St. Leonards. 
-136 
-148 
+ 12 
+ 331 
Salisbury. 
- 1 
- 83 
+ 82 
±78/ 
Tunbridge Wells 
-186 
- 97 
- 89 
E 
Armagh. 
-139 
- 31 
-108 
Einiiskillen. 
- 43 
+ 37 
- 80 
- 611 
Sligo. 
- 55 
-127 
+ 72 
±97/ 
Carrick-on-Shannon . 
-140 
- 12 
-128 
F 
Swansea. 
+ 26 
+ 31 
- 5 
lid : 
Cardiff. 
— 62 
- 32 
- 30 
Milibrd. 
- 48 
- 51 
+ 3 
Ct 
Charleville. 
-116 
-102 
- 14 
Killarney. 
+ 1 
- 3 
+ 4 
+ 351 
Bantry. 
+ 57 
- 39 
+ 96 
± 42 / 
Valentia. 
0 
- 55 
+ 54 
1 
' 
We take it that the result will be considered satisfactory if the mean difference, 
calculated irrespective of sign, is of the same order as the error of experiment, and if 
the algebraical mean is considerably less than this quantity in each district. 
The first result would show that the differences are not unreasonably large, the 
second, that errors of a particular sign do not predominate in any locality. 
In our “ 1890 Memoir” we proved that the probable differences between two 
measurements of the Horizontal Force and Dip v/ere 0'00056 metric unit and 
F‘0 respectively. 
The first of these numbers was calculated from the differences between observations 
made at the same time, and, as we pointed out, did not include the error of determi¬ 
nation of the temperature. 
4 a 2 
