REPORT OF THE KEW COMMITTEE. Ixiii' 



With, regard to date, each curve when taken off the cylinder should have 

 both the day of the week and of the month written upon it, and when it 

 reaches Kew it ought to be inspected by the assistant there in order to see 

 that the observer has attached the proper day of the month alongside the 

 day of the week. 



He should also see that the week's curves sent are dated consecutively. 



With regard to time, a prick made in the small time-scale of the metallic 

 sheet denotes in terms of the hour-lines ruled on this sheet, the moment of 

 starting, and a similar prick that of taking off. These pricks ought to denote 

 the true chronometer times of starting and taking off very nearly, if the in- 

 strumental clock has been properly regulated. All stoppac/es of the instru- 

 mental doclx: ought to be marked. 



It ought also to bo noticed that the cylinder is well clamped, otherwise the 

 friction of the pencil upon the cylinder may occasionally overcome that of 

 the clamp, in which case the cylinder will slip. 



With regard to errors of tabulation, the assistant at Kew ought in the first 

 place to ascertain that the curve is tabulated under its proper date. Probably 

 an intelligent inspection of the direction- and velocity-records in connexion 

 with the tabulated results wiU be sufficient to determine this point. 



A simple system of subsidiary tahidations has been adopted in order to 

 check the dii-ection-results. The observer at the outlying observatory is 

 requested to write down on a separate sheet in numbers the direction of the 

 wind at each hour as read from the curve by his eye, and compare it, as in 

 the case of the Barograph and Thermograph, with the tabulated results. The 

 differences between the two columns or A — B ought to be inspected at Kcav 

 and when they are greater than tivo points the ease ought to be examined' 

 and any error detected ought to be corrected at once. With respect to direc- 

 tion, fractional parts of a point ought not to be recorded. 



_ With regard to velocity-traces, the action of the instrument is sucli as to 

 give by a glance at a curve the whole space travelled over by the wind for 

 that day. Perhaps, therefore, it will be a sufficient check upon the velocity- 

 records if, in addition to an intelligent comparison of the traces and tabula- 

 tions, each day's results are added up and the sum total compared with that 

 derived by glancing at the curve. When the difference between these two 

 daily sum totals is greater than one-twentieth of the whole, the tabulated velo- 

 cities for that day ought to be gone over again, and if any error is detected 

 it ought to be put right at once. 



It is probably unnecessaiy to check the recorded oscillations, as these are 

 of inferior scientific value, and additional labour bestowed upon them would 

 appear to be superfluous. 



Finally, in order to keep a check upon small errors, the system of making 

 at Kew forty remeasurements for each month, both for direction and velocity, 

 has been adopted. 



The following Table exhibits the results of the method employed for testing 

 the accuracy of the Anemograph tabulations as regards small errors. 



It will be seen from this Table that the standard of accuracy as repre- 

 sented by the smallness of the mean monthly differences has gradually in- 

 creased np to the end of the year. 



