224 REPORT — 1895. 



The following data are extracted from tlie annual Kew ' Reports ; — • 



Table IX. 



The mean values thence deduced for the secular variation are a 

 decrease annually of 6'"9 in declination, and an increase annually of 

 10"'^xl95 C.G.S. units in horizontal force. It will be assumed that 

 these secular variations proceed uniformly throughout the mean year, got 

 by combining the five years 1890-94. To eliminate the secular variation 

 one adds to the observed values +0'-575t in the case of the declination, 

 and — 10"^xl"625< in the case of the horizontal force, t being the time 

 ■elapsed in months from the middle of the mean year. Subtracting the 

 mean value for the year from the monthly means thus corrected, one 

 obtains the annual inequality. 



§ 17. To make the true position of affairs clear, a brief explanation is 

 necessary of how the magnetic curves are standardised at Kew. Of late 

 years the practice has been to determine the value of the zero line for 

 each month's curves by reference solely to the absolute observations of 

 that month. The same instruments are used for each observation of an 

 element, and every observation is independent of the others, except that 

 the constant usually called P in the formula for the horizontal force — 

 i.e., the coefficient of the secondary term in the exjiression for the deflect- 

 ing force — is determined from a whole year's observations. 



One of the most probable and subtle causes of error one has to provide 

 against in getting out an annual variation of any physical phenomenon is 

 <a possible secular or annual variation in the measuring instruments. 

 This is especially the case with apparatus sensitive to changes of temper- 

 ature. Now it is unquestionably true that the horizontal force magneto- 

 graph is affected by changes of temperature, and though the underground 

 chamber it is worked in at Kew has a very small diurnal variation of 

 temperature, it has a considerable annual fluctuation. Thus, however 

 carefully temjierature corrections might be determined and applied, the 

 suspicion of an ' annual inequality ' being far other than it seemed might 

 not unreasonably be entertained, if the ultimate reference were to the mag- 

 netograph curves, either unstandardised, or standardised by reference to 

 tbe mean of a year's absolute observations. It is partly to provide against 

 this that each month's curves are referred to that one month's absolute 

 observations. 



These observations are taken about once a week and scattered over 

 the month, so that any secular or annual variation in the magnetographs 

 themselves must be very nearly eliminated. 



As regards the absolute instruments there is, so to speak, no higher 

 court of appeal. There is no obvious ground for suspicion. The hori- 

 zontal force magnet has a temperature correction to apply, but this is only 

 to allow for the difference in its temperatures at the times of the vibra- 

 tion and deflection experiments in the same observation. This difference 

 in temperature is very small and very irregular, and even if the tempera- 



