106 Transactions of the Royal Society of South Africa. 
Taking Table II as it stands, we have a maximum in each month about 
two hours after sunset. Now so-called sheet lightning is generally regarded 
as the reflection of distant thunderstorms. There is much to be said for the 
idea. On the other hand, since thunderstorms begin to decline in frequency 
before sunset, one would hence have expected, in view of the shortness of 
Kimberley twilight, the maximum to fall a little earlier than it appears to 
do. We should, indeed, anticipate a maximum as soon as the fading twilight 
glow would allow the lightning to be easily seen. And for more reason, 
that my sunshine charts are changed every day during civil twilight, and 
advantage is taken of the opportunity to search the horizon for lightning. 
Another curious feature in the table is the greater relative rapidity with 
which the frequency of lightning runs down after about 8 p.m. as compared 
with the frequency of thunderstorms. Such a falling off depends in the 
main, no doubt, on the circumstance that Kimberley thunderstorms travel 
generally west to east. 
Tables I and II are combined together in Table III. The total 
duration is 7130-5 hours; i.e. rather less than 1 hour in 29, on the 
average, has lightning or thunder or both together. In January the fre- 
quency rises to nearly 1 hour in 15. 
At the foot of Table I is given the number of nil months, that is to say 
the number of months without thunderstorms in the period of twenty-three 
years. It may be gathered from this that a month without some sort of 
a thunderstorm is not to be expected in the summer half-year, October to 
March. The winter half had thirty-nine thunderstorms between them, prac- 
tically every alternate July being free. 
The next two lines in Table I show the greatest and least frequency in 
hours of thunderstorm weather in any one month of the period. March 
heads thelist with 52*5 hours (in 1908), never falling below 7"3 hours (in 1902). 
No June, on the other hand, ever had more than 6 hours. 
The next line shows the percentage of thunderstorm to the whole electric 
disturbance observed, as given in Table III. The great peculiarity here is 
the sudden rise in the thunderstorm ratio from May to June. It may be 
suggested as a fair inference either that the early winter thunderstorms 
overspread a greater area of country than the late autumn ones do ; or (less 
likely) that they are local squalls which especially favour Kimberley. The 
local character of summer thunderstorms is well known. One of the most 
curious in my own recollection was a solitary large cloud, distant to the 
south-west of Kimberley, from three points of which rain, accompanied by 
lightning, was falling.* 
The last line shows, in modern notation, the time of sunset for the 
* C/. an account in the ' Meteorological Magazine ' for March, 1920,quoted from ' The 
Times,' of Central African thunderstorms observed from mid-air : " I have seen as 
many as a dozen of these storms [? at once] within an area of 100 square miles." 
