74 Transactions of the Boyal Society of South Africa, 
with a thin brass collar, which served to keep the float rod vertical while 
effecting little frictional retardation of its vertical movements. When 
fixing the machine in position over the borehole care was taken to level 
the drum with a spirit-level, so as to ensure that its axis of rotation should 
be truly vertical. 
As Messrs. Eayner and Koberts use the water of the boreholes for 
irrigation, and frequently have to regulate the flow to suit their require- 
ments, it is only at certain limited periods that opportunities of using the 
recording apparatus occur. 
The apparatus was fixed up over the small tank A at borehole No. V., 
and records were taken when opportunities occurred. The first long 
periods for which I obtained uninterrupted records were (1) the week 
May 22 to May 29, and (2) the fortnight June 4 to June 18, 1905. 
During the taking of these records all the boreholes were closed except 
No. v., except for three occasions, on each of which No. IV. was opened 
temporarily for a few minutes. The effects of these temporary inter- 
ferences are perfectly defined on the records, and make no trouble, as the 
normal course of the curve can be easily and satisfactorily interpolated 
owing to the very short duration of the perturbations. 
Diagram 4 shows a tracing of the record sheet for the 24 hours' period 
8 a.m. June 12 to 8 a.m. June 13. This record, besides showing the notch 
effect of one of the three interferences above mentioned, shows the general 
characteristics common to all the day records of the series. (See also 
curve A in Diagram 9.) 
The most striking general feature of the curves obtained is their 
wonderful regularity. Each day record shows two maxima and two 
minima, and these turning-points occur at very regular and approximately 
equal intervals, as estimated by general inspection. By determining the 
times of the first and last maxima in the May week and dividing the time 
interval by the number of complete wave-lengths on the record, I quickly 
obtained a first approximate estimate of the average wave period as about 
12^ hours. By a similar process the records for the June fortnight gave 
the same result. 
On one side of the curve will be observed a large number of nearly 
vertical "hair" lines. Each of these "hair" lines records the escape of 
an unusually large gas-bubble, which on its arrival at the top of the 
borehole has impinged on the under surface of the float and caused a 
momentary upward jolt. 
It is worthy of notice that in general the occurrence of these large 
gas-bubbles is much more frequent near the periods of high water than 
about low-water periods. 
Another general feature of the records is the occurrence of a frill of 
minor fluctuations, with period varying from 5 to 20 minutes, and of an 
