62 



REPORTS ON THE STATE OF SCIENCE. — 1915. 



measured with care and discussed at length, but a brief summary will 

 suifice here : 



May 4-10. The mean temperature first rose and then fell. But 

 it will be convenient to consider first the mean diurnal inequality which 



came out in degrees Fahrenheit, 



l°-4 cos (9 



h. 



20-8) 



6 being expressed in hours. The late hour for the maximum, nearly 

 9 o'clock in the evening, raises hopes that we may be able to separate 

 the effects of internal and external temperature. The first harmonics 

 for the two machines were 



mm. h. 



Milne-Shaw 24-2 cos (6 - 180) 

 Milne-Burgess 5-6 cos (0 — 20-3) 



The former measures and coejficients were inadvertently expressed 

 in units of 0-01 in. ; converted into mm., the former mean values are 

 16-8 and 3-8. These are smaller than the 24-2 and .5-6 now found, but 

 the ratios, 44 and 43, are very nearly the same. 



The phases, however, do not differ by nearly so large an amount. 

 M.-B. follows M.-S. by 2-3 h. only instead of by 6 h. as found in Table V. 

 It will be seen, moreover, that while the maximum of M. B. falls near 

 that of the thermograph, M.-S. precedes the thermograph by a large 

 interval — nearly 3 hours. Now, if the disturbance of the instruments 

 is due to some temperature effect outside the Observatory — tilt in the 

 vallev, for instance — the time of maximum would be different from that 

 of the thermograph. For instance, if it depended on the Sun's altitude, 

 the maximum should fall at noon. If the effect is a composite one, the 

 maximum would fall somewhere between noon and 20-8 h. (internal 

 maximum), as it does in fact. We have thus a presumption of a com- 

 posite character. 



The presumption is strengthened by the magnitude of the coefficients ; 

 —1° of temperature corresponds to 4-0 mm. for M.-B. and actually 17-7 mm. 

 for M.-S. It seems unlikely that these movements, especially the 

 latter, can be due directly to the 1° change in internal temperature. It 

 seems much more likely that they are due to the miich larger external 

 changes, of which the 1° internal change is only a fraction. 



Further evidence in this direction is afforded by the changes of the 

 mean from day to day, which can be formed when the diurnal change 

 is eliminated. These were formed for every available day in the two 

 periods May 4-10 and May 31-June 12, and it will sufi&ce to give the 

 mean results : 



