6 
HARRINGTON. 
storms then decrease to the veranillo, at which time they may 
cease entirely for a few days or weeks. They begin again 
after it and reach a second maximum in October, when they 
gradually decrease until December. The rainfall phenom¬ 
ena, though short and intense, are very local in area and 
have marked limitations in time. Their intensity is largely 
a question of topography and wind. Hence the most 
marked and striking variations exist, similar to those of 
local storms elsewhere, but more marked in character. 
Hence also marked localization of rainfall. A few rods 
will sometimes make one pass from within a torrential rain 
to a spot without rain. Hence also remarkable variations 
in the rainfall from year to year or in the same month 
of successive years or in the monthly rainfall at adjacent 
stations. 
3. The Annual Rainfall .—The observations on which we 
have to rely are rarely synchronous, and consist chiefly of a 
few years’ observations at each station, but scattered at the 
various stations through half a century. This makes it 
necessary to compare the longer series and see if there is 
any periodicity to be found of such character as to render it 
necessary to adjust the observations to the same date. Such 
a comparison is given in Table II for the ten or (combining 
Panama, Naos, and Taboga) eight best series, and a succes¬ 
sion of maxima and minima at once appear, the ratio be¬ 
tween the two being 4 to 7, 5 to 9, and even greater in some 
cases. It appears that the times of maxima are practically 
the same throughout Central America, and are as follows: 
Maximum. 
1856.. . 
1861.. . 
1866-7 
1872 .. 
1878-9. 
1886-7 
1893... 
Interval of years. 
5 
8 
6 * 
These show a fairly regular periodicity of 5 to 8 years, 
not regular enough to justify discussion in such a fragmen¬ 
tary series of observations, but regular and large enough to 
