376 



Mr. F. Chambers. Xote. 



[Nov. 20 r 



y-series corresponds to " quartile " of the ,r-series. For, by the 

 ordinary theory 



h (probable error) = -4769 . . . 



.-. =Ji (log Q— log a) 

 . • . log a+ prob. error =log Q. 



In words — the term which differs from the Arithmetic Mean by 

 probable error in the y-sevies is the log. of the quartile in the ^-series. 



We infer that no new tables are necessary for the practical working 

 of our method. We require only the tables of the error-function and 

 of hyperbolic logarithms. Of course common logarithms may be 

 used, if we remember to introduce the appropriate modulus into our 

 formulae. 



In conclusion, I desire to acknowledge my indebtedness to Mr. 

 Galton, not only for the suggestion of the problem I have here 

 attempted to solve, but also for many valuable practical hints in the 

 working. 



XIY. 



(Received September 19, 1879.) 



Colaba, Bombay, 



29th August, 1879. 



Dear Sir, 



I regret that in my paper on the " Diurnal Variations of the Wind 

 and Barometric Pressure," published in vol. xxv, p. 402, of the " Pro- 

 ceedings of the Royal Society," I have inadvertently referred the 

 Bermuda wind observations to civil time instead of to astronomical 

 time. 



The correction of this error shows that the variation of the values 

 of the hourly inequalities of the wind components at Bermuda is very 

 much less during the honrs when the sun is above the horizon than 

 during those when he is below it, especially in the winter half year. 



The calculated barometric variations, depending on the wind varia- 

 tions, should also be referred to astronomical instead of to civil time ; 

 but the general reasoning of the paper remains the same, leaving the 

 principal portions of the variations (viz., those which have half a day 

 for their period) unaffected. 



(Signed) F. CHAMBERS. 



The Secretary of the Royal Society. 



