16 REPORT— 1866. 



tioiis of winds are from the N. on 80-1 days, from the E. on 72-3 days, from the S. 

 on 887 days, and from the W. on 123-7 days. The force of wind was estimated 

 from to (3; being a cahn and 6 a hmiicane, the mean force heing 0-Q; but 

 Robinson's anemometer is not stationary more than twice a year. 



During a residence of nearly ten years in the camp the author has had the good 

 fortune to observe that, in comparison with any other station, civil or military, its 

 "bill of health " forms a striking contrast to their published statistics. 



On the Method adojpted at Utrecht in discussing Meteorologiccd Observations. 



By Dr. Btjys-Ballot. 



On an Error in the usual method of ohtaimng Meteorological Statistics. 

 By Peancis Galton, F.B.S., F.O.S. 



The meteorological statistics of the ocean have been hitherto obtained by ex- 

 tracting observations from the logs of different ships, and by sorting those that 

 were made in ditferent geographical divisions of the ocean into corresponding 

 groups. The usual geographical divisions are bounded by each 5th degree of 

 latitude and longitude, and they therefore are 300 miles in length, and have an 

 average breadth of loO miles. Each of the "roups is treated as if it were composed 

 of observations taken at irregular periods, by a single person stationed at a hxed 

 observatory in the centre of the group, that is to say, the barometer, thermometer, 

 and other elements are determined by computing the simple mean of all the re- 

 corded observations. The propoi-tion of winds that IjIow from the diflerent points 

 of the compass is computed in a similar manner. Only one limitation is exacted in 

 respect to the admission of an observation into a group. It is that it should not 

 have been made at an interval of less than eight hours from any other observation 

 by the same ship, already included in the same group. Were it not for this limi- 

 tation, a zealous observer might contribute hourly, or j'et more frequent observa- 

 tions, which, by their multitude, would prevent the scantier observations of other 

 ships from exercising a just influence on the a\'erage. In an extreme case of this 

 description, the weather met with by a single ship on one particular voyage might 

 mainly govern the computed results. 



In a recent report on the condition of the Meteorological Department of the 

 Board of Trade by ^Ir. Farrer, Captain Evans, and himself, were pointed out many 

 objections to the existing methods of computing ocean statistics. The object of 

 the present paper is to draw attention to yet another objection, and to show that 

 an additional limitation is required before an observation ought to be admitted 

 into a group. The objection was, that the observations made by a sailing ship are 

 more numerous in respect to antagonistic winds or calms than in respect to 

 favourable weather. Therefore, as some parts of the ocean are mainly frequented 

 by outward-bound and others by homeward-bound ships, the means of the re- 

 corded observations in those parts must differ materially from the true average 

 weather. When favourable "u-inds are blowing, a ship is rapidly wafted across the 

 area of observation, and comparatively few observations are made within it. The 

 wind may continue blowing, but the ship is unable to record its continuance after 

 it has left the area in question. On the other hand, if an antagonistic wind blows, 

 or if calms or light breezes prevail, then the ship is delayed within the area, and 

 continues making observations during the whole, or nearly the whole period 

 of their continuance. Taking one course with another, a ship sailing -with a 

 favourable wind crosses one of the usual five-degree divisions of the ocean in twenty- 

 four hours, or, in other words, in three eight-hourly periods of observation. There- 

 fore the observations made by a ship resemble observations made at a fixed obser- 

 vatory under the instructions that only three eight-hourly observations were to 

 be taken din-ing the continuance of winds, say, from the northerly quadrant, but 

 that when the wind was in the southerly quadrant the observations were to be 

 continued dming the whole of its duration. No one would be inclined to accept 

 the means of these observations as a just statement of the weather, yet this is 

 precisely what is given by the method of compilation adopted by the Meteorolo- 



