54 1.8.C. Proceedings of the Ninth [N.S., XVI]. 
kilometres above sea-level and various other elements. In the present 
r it is shown that —— exist other heights at which the pressures 
i e is also true for 
The position of 9 km. ag es not unique, and from the statistical point 
of view the peculiar tS oetanns of 9 km. is not sustained. 
the correction of a coefficient of correlation for 
observational errors.—By P. C. MAHALANOBIS 
Chapman (in a paper pe in the Proc. Roy. Soc. eo. ue 
dianwabect ie effect of correcting for ‘Hacc vanssindl errors. 
obtained very high values for die corrected et air coreieae ona 
eéachin ng unity or absolute certainty in certain 
the present paper it is shown that se ‘eee ciate statistical assu 
tions are involved in the above beet ihe also that these statistical 
assumptions are - justified in the cases of the upper air data consi- 
dered by Chapma 
On the probable error of the component frequency con- 
stants of a dissected frequency curve.—By P. C. MAna- 
LANOBIS 
ae Syms metric Triage of the first type.— Pearson has giv 
the soma for the dissection of a symmetric curve oO latybuctis 
curves, the component ra are different, while the standard deviation 
ains 
me, res oe 
ent m are obtained and the results used for the determination of 
the reliability of dissection of Ra frequency distribution of heights of 
different Bengal Castes and Tri 
Part II. Symmetric ciao on of the second type.—For leptokurtic 
curves the standard bags vi _ ms are agp while the means are the 
Same. Probable errors are found and applied to a numerical case of 
dissection of Alagio ndini acne 
On the probable error of ase Serta by linear 
interpolation.—By P. C. ManaLaNno 
n Physics linear tarpon is very common on used for finding 
the ‘‘ best” value of a constant. The straight line used for interpolation 
should of course, be the line of best fit, which is statistically the same 
the principal axis of the correlation ellipse sent paper the 
probable er a constant interpolat m the line of best fit is 
calculated in terms of the means, the standard deviations seh the 
correlation of the two variables. Namerical examples will be give 
An automatic “ make and break” key for the heating and 
aes potential circuits of a Cooledge X-ray tube.—By 
P. Harrison and NaRENDRANATH SEN 
On the experimental fae oe of the Temperature 
Radiation of Gases. —By HA 
uthor described an experiment = h he carried out at the 
Laboratory of Prof. W. Nernst of Berlin ae rae experimental de- 
onstration of the temperature radiati ie gase he tion was 
previously attacked by J. J. Thomson (Phil. Mag. 1905, 6, | 
M (Proc. Roy. Soc. Lond., Vol. 93) but their results were 
omson worked with aay 8 at acLen- 
nan with the vapours of Hg, Cd, Zn i unsen flame. The author 
showed that the non-success. of their exierinacts was sas to the fact 
