PERIODICITY OF GOOD AND BAD SEASONS. 73. 
made. At the time it was not seen what a great saving of trouble 
and time this very simple device would ensure, but it is abun- 
dantly evident now, that this diagram has shortened by some 
years the attainment of the object I had in view. 
As a preliminary to making the diagram, the particulars of the 
weather in this Colony from all sources, for every year of our 
history, were carefully examined, and the years simply classed as 
good or bad, that is, having sufficient or insufficient rainfall; a 
form was then prepared with a vertical space for each year, and 
across these a zero line was drawn to divide the good from the 
bad, and beginning with 1895 I filled in for that year, and below 
the line a convenient length of the column in red ink, the length 
was simply to catch the eye ; then for 1894, a good year, I filled 
in with black ink above the line a space equal to the red in the 
vertical space for 1895. The two years were thus contrasted, 
simply as good and bad; the question of how good or how bad 
was purposely left out. The diagram was then completed, each 
year being treated in the same way back to 1788. It was at once 
apparent that the drought which has been lettered A for conveni- 
ence was the most regular in its recurrence, and the most exten- 
sive in time, lasting as it does from three to seven years. 
A vertical red line was then drawn through A between the first 
and second years, and it was found that the interval was regular 
and exactly nineteen years. In giving the date of any drought, 
the year after this line was used, being in all ordinary cases the 
middle year. Drought D was then examined and found to recur 
with nearly the same regularity, and a short vertical line was 
drawn under it to mark the point which has been taken as the 
centre, and this also recurs at intervals of nineteen years ; there 
are three others, B, C, and E which also recur with an average 
interval of nineteen years, but they are not quite regular, may 
in fact differ from time to time a year or even two years, this 
uncertainty of time is indicated in the diagram by making cost 
red space shorter. 
