196 FLORIDA GEOLOGICAL SURVEY-I3TH ANNUAL REPORT

294 days at Ocala (and doubtless still less a little farther north) to practically 365 on Merritt's Island and south of there, where often a whole year passes without frost. (The imaginary "frost line" lies considerably farther south, however, for there is probably no place in Florida, with the possible exception of the Keys, that has not had frost at some time within the memory of persons now living.)
 Although it is not feasible to present figures on that point, the temperature of course varies from year to year, and some of these variations have made the difference between success and failure for those who are always trying to raise tender crops as far north as possible. The severe freezes of 1895 and 1899 almost wiped out the orange industry in Florida (which was then largely concentrated in the latitude of Ocala and farther north), but since then many of the larger groves have been established farther south, and more attention has been paid to locating them on high points oi near lakes for protection from frost, and installing heating devices to use at critical periods, and there has been comparatively little trouble from that source in the last twenty years. The lake region has an advantage over most of the others in its abundance of hills and lakes.
 Snow is of course practically unknown. The extreme variation in rainfall from place to place are not great, but the lake region seems to be a little the driest, perhaps because farthest from the coast. Although there may be considerable variation from year to year, it is hardly enough to cause any serious inconvenience, for there is nearly always enough rain to prevent drought, and at the same time the topography and soil make floods almost impossible. On the sandy uplands the heaViest rain sinks into the ground almost immediately, to appear gradually later in swamps and springs.
 Over half the rain falls in the four warmest months, and over two thirds in the six warmest months, thus counterbalancing evaporation to a large extent and keeping the level of lakes and streams more constant than in most other parts of the United States. A slight correlation can be noticed between the late summer precipitation excess and soil fertility, the excess being less in the hammock belts than in the lake region and flatwoods; as if the soil itself