xxii Transactions of the Royal Society of South Africa. 
quantities of natural inflammable gas (methane), while sulphuretted 
hydrogen is present in notable quantities in solution in the water. 
Measurements of the pressure at which the water issues show a 
remarkable fluctuation, in many respects analogous to the tidal fluctuations 
of the sea. 
A series of direct measurements, covering several days, established the 
fact that there was a real fluctuation, both in the amount of water dis- 
charged and in the well-pressure. Continuous records were then obtained 
over longer periods, by means of clock-driven, self-recording apparatus, in 
order to study the precise nature of the fluctuations. 
The longest continuous record obtained extends over a period of fifteen 
weeks. This graphical record shows that the semi-diurnal fluctuations 
attain a maximum amplitude at fortnightly intervals at times correspond- 
ing to the times of New Moon and Full Moon throughout the fifteen weeks' 
period. 
This record further demonstrates the fact that the mean daily water 
pressure rises with each fall of barometric pressure, and falls with each 
rise in barometric pressure as recorded concurrently at the farm by means 
of a barograph instrument. The time scale on this fifteen-week record is 
about 11 inches per week. 
Eecords obtained for shorter periods on a time scale of 13^ inches per 
day were found to be much more suited for detailed critical examination 
and analysis. In particular, the record for a certain fortnight during 
which the barometric pressure was very steady (and, consequently, its 
interfering effect almost negligible), was selected. The times of all the 
turning-points were carefully determined in terms of S.A. Oflicial time. 
The heights of all the turning-points of the curve were also determined 
in inches. 
Similarly, the co-ordinates of all the turning-points for that fortnight 
were determined on the tide gauge records of the S.A. ports of Cape Town, 
Port Elizabeth, East London, and Durban. 
For general comparison all these measured data were plotted in parallel 
lines on the same time scale, and the general resemblance of the Well 
curve to the curves of the coastal tide records demonstrated. 
The original Tarka Bridge Record for the fortnight was then subjected 
to a process of harmonic analysis for the purpose of determining the 
periods of the principal harmonic components of the curve. The particular 
method used was described by Chrystal as the method of " Eesiduation " 
(Trans. Roy. Soc, Edinburgh, vol. xlv., part 2, pages 385-387). 
This method involves no assumptions as to the causes operating to 
produce the curve. One by one the various simple harmonic components 
are disentangled from the compound curve with their periods unaltered, 
but with their amplitudes considerably reduced. 
