90 
Transactions of the Boyal Society of South Africa. 
(iii) From 2nd L.W. to 26th L.W. was a period of 299 hrs. 33 min. 
^ 299 hrs. 33 min. 
•*• J-D = 24 ^ ^9 
Averaging these three values, we get as a first approximation — 
Td = 12 hrs. 27 min. 
This is of course really a second approximation to the value of T^. 
Again, returning to curve A and residuating it with respect to 
T = 12 hrs. 27 min., I obtained a residual curve F almost identical with 
the curve C already obtained. 
On curve F the following results were obtained : — 
(i) From 1st L.W. to 14th L.W. was a period of 311 hrs. 34 min. 
311 hrs. 34 min. 
.-. Tj. = = 23 hrs. 58 mm. 
(ii) From 2nd H.W. to 14th H.W. was a period of 286 hrs. 52 min. 
286 hrs. 52 min. , . 
.-. Tp = ^2 ^ 23 hrs. 54 mm. 
(iii) From 3rd H.W. to 14th H.W. was a period of 263 hrs. 50 min. 
263 hrs. 5 0 min. _ . 
.-. Tp = Yi = 23 hrs. 59 mm. 
Averaging these three values we obtain — 
Tf = 23 hrs. 57 min. 
This is of course really a second approximation to the value of Tq, 
and it thus appears that the limits of working accuracy were attained in 
the first residuation. 
The curve F was next residuated with respect to its dominant harmonic 
period T = 23 hrs. 57 min., and yielded a curve H. No regular periodic 
undulation can be discerned on the curve H. By placing the eye near the 
level of the tracing cloth and glancing along the foreshortened curve, one 
distinguishes a fluctuation of irregular periodicity. Three crests and two 
troughs can be seen in the entire curve. As the curve is very flat and the 
undulation very long the positions of some of these turning-points can 
only be estimated with a probable error that might amount to nearly a 
day, while in the case of others the error might not exceed 6 hours. 
The curve begins with a slight upward movement reaching a maximum 
