220 



KNOTT AND TANAKADATE 



observîition in 1883. In the circumstances we may regard tlie con- 

 ditions of the diurnal variation as essentially the same in the years 

 1883 and 1887. 



It will be noticed that the curves obtained at Wakwan (No. 62) 

 and Meho ( No. 63 ), two of the Korean Stations, have peculiarly large 

 ranges — nearly 11' in both cases. The two curves are so similar and 

 so smooth that it is difHcult to believe their exceptional character to 

 be other than a real thing. It will be further noticed, however, that 

 during the nnrnth of August generally the diurnal range is distinctly 

 greater than the mean just given. We may, indeed, by grouping the 

 different sets of observations according t(j uKKiths, obtain an indica- 

 tion of the annual chano-c of range. These monthly means of diurnal 

 rangfe are as follows : — 



Even if we were to throw out the two excessive Korean values, 

 the Auo'ust mean W(mld still be distinctlv higher than the Julv or 

 Septeml)er mean, being in fact 8'. 8. All this is quite in accordance 

 with well established facts.* 



There is one other feature presented by Mr. Tanakadate's curves, 

 which seems worthy of remark. It is that the hour of maxinuim 

 deviaticm appears to come distinctly earlier in the later curves. This 

 is espacially well marked in the curves on Plate XIV. At Hagi, 

 Hamada, Matsue, indeed, the maximum seems to fall at noon instead 



* Compare Table VI. in Art. Meteorologt, (Section Terrestrial Magnetism), Encyclo- 

 paedia Britannica (Ninth Edition.) 



