454 



In addition to data tor single months of the year the table gives diurnal inequalities for the year as a 

 whole, and for midwinter (May, June, July), equinox (March, April, September, October), and mid 

 summer (November, December, January). Midwinter was limited to three months, so as to include only 

 days in which the sun never rose, and a corresponding limitation was, made to midsummer. In the three 

 midsummer months the inequality for the month is based on only a single month's observations ; in ihr 

 other nine months the results are means from two years. 



In the table " Itange " denotes the difference between the algebraically greatest and least of the mean 

 2-hourly values. 



'2. The data recorded in Table I. were employed for the calculation of Fourier coefficients. Even at a 

 first class European station with hourly readings, two years is too short a period to furnish thoroughly 

 representative values for Fourier coefficients, especially those for periods shorter than 12 hours. In the 

 Antarctic the regular diurnal inequality is so small, and the irregular changes so large, that 2-hourly 

 readings from two years seemed too slender a basis for the calculation of more than the 24- and 12-hour 

 terms. 



In Table II. ^ and :, denote the amplitudes of the 24- and 12-hour terms, unity representing 1" F. ; aj 

 and a. are the corresponding phase angles. The series would run 



O sin (/ + a-i) + c.-> sin (21 + a-i) + ..., 

 t being counted from (local) midnight, and 1 hour being taken as equivalent to 15. 



TABLE II. Fourier Coefficients, Diurnal Inequality. 



111 considering the results, it must be remembered that the existence of finite values for r\ and <- is not 

 by itself sufficient proof of the reality of a diurnal inequality. If, however, no real diurnal inequality 

 existed, the values of c^ and e., would most probably fluctuate in an erratic fashion from month to month, 

 while the values of a.\ and &> would be quite irregular. 



As compared to data for a temperate station, the phase angles for individual months in Table II. do 

 unquestionably show large variations, and it can hardly be claimed that these variations are wholly free 

 from quasi-aecidental causes. In the case of a.\ the values for April, July,* and August appear outstanding, 

 but they agree closely amongst themselves. 



Taking the seasons, the value of a,\ for midwinter differs sensibly from that for equinox, but the 

 difference is in the same direction as that of equinox from midsummer, and is not so very much greater. 

 In these three seasons the positive crest of the 24-hour wave is reached approximately at the hours 

 2h. 50m., 2h. Om., and Oh. 35m. in the afternoon, being latest at midsummer. In the mean diurnal 

 inequality for the year the crest of the 24-hour wave conies approximately at 2h. 20m. p.m. 



* The angles 4' and 304", it should be noticed, arc equivalent. 



