Ix REPORT — 1869. 



Errors in Trace and TahvJation. 



The arrangement proposed for ensuring the entry under the proper da*e 

 into the tabulation forms of the measuremeuts of the Thermograph curv/s, 

 and of the Standard readings corrected, is almost precisely the same as f.iat 

 stated in the case of the Barograph. 



Having ascertained the entry into the tabulation forms under their propt" 

 dates of the tabulations, and of the Standard readings corrected, -we come in 

 the next place to consider the check upon accuracy of tabulation, and here, 

 as in the case of the Barograph, it will be necessary to consider separately 

 large and small errors. 



In the first place, with respect to large errors, in order to prevent entirely 

 their occurrence, it is necessary to resort to the system of subsidiary tabula- 

 tions. An instrument for this purpose has been devised at Kew. It is un- 

 necessary here to state its principle of construction ; suffice it to say, that 

 the results furnished by it arc used in the same manner as in the case of the 

 Barograph ivory scale already mentioned. By this means correct columns 

 of tabulated readings may be obtained. Again, Avith regard to the Standard 

 readings, all that appears to be necessary is to examine both the accuracy of 

 entry of the Standard reading coi-rectcd, and the accuracy of tabulation for 

 aU those cases in which the recorded Thermograph temperature is more than 

 half a degree diiferent from the corresponding Standard reading, and to make 

 any correction that may be found to be necessary. When a Standard read- 

 ing is evidently wrong, it ought to be noted as such on the curve, and should 

 not be made iise of in calculating the monthly mean difference between 

 Standard and Thermograph readings. Before leaving this subject, it ought 

 to be stated that the tabulating instrument as well as the subsidiary scale, 

 are both so arranged as to ensure reading the proper point of the curve for 

 every odd hour. 



It ought to be noted that, in tabulating from the Thermograph curves, the 

 tabulating instrument should be set from those observation hours where 

 there is little thermoraetric fluctuation. 



AU the dry-hidb readings ought to he compared with the corresponding ivet- 

 hulb ones, and should the latter ever appear higher than the former, the case 

 ought to be marked. 



The maximum and minimum temperatures furnished by the outlying ob- 

 servatories ought to be checked. 



All Ia)-ge errors may, it is hoped, be completely obviated by the means now 

 described. 



With regard to small errors, the plan adopted is the same as that for the 

 Barograph, viz. : — ■ 



(1) To record the monthly mean difference between the sinraltancous 



Standard and Thermograph readings. 



(2) To make forty remeasurements from each month's curves at Kew. 



The following Table exhibits the results of the method employed for test- 

 ing the accuracy of the Thermograph tabulations as regards small errors : — 



