19 
heads (fig. 24). The roots of Aclirosticham are fleshy, with the cortex 
composed mostly of iiitercelhiliir spaces. An occasional root, Avithout 
structure which is evidently different, grows upward out of the ground, 
SAVANNA-WOOU. 
* 
L 
This formation includes very nearly the whole of the coastal plain. 
Here, at and near the sea level, the dry season is longest and driest and 
the hills give no protection against the winds which, though that far 
from as strong as they would be farther from the equator, suffice con- 
stantly to remove the moister air next the groiuid and vegetation. 
More than one-third of this plain lias been made drier artificially by 
clearing for cultivation, and there are areas of many hectares which seem 
naturally to be in grass. The most conspicuous features of this formation 
are the moderate size of the trees and their very open stand ; still, in the 
back part of the i^tlain are sc\'cral rather close groves of some enormous 
Mijrlaceous species, believed to be a Eucalyptus.^ In spite of the lofty 
trees and close stand this portion is bionomically, unmistakably a part 
of the savanna-wood and not of the high forest. With regard to the 
ferns, the notal)le feature of this formation is the absolute absence of 
any characteristic epiphyte. Asplerdum musaef oil urn , Drynaria querci- 
folia^ and a few other species of the high forest are occasionally to be 
found, but no single one can be regarded as at home here. The com- 
monest of these, D. qnercifolia, is almost always deciduous in the savanna- 
wood, being quite without green leaves during the driest months, but in 
the high forest it is very rarely found in this condition. 
I made hygrometric determinations^ on the strand and about 300 meters 
inland, three times daily, for a long enough period to get a very good idea of 
the comparative atmospheric dryness of the two places during the dry season. 
Tliese aje publisJied in detail a.s an appendix to my paper on the coconut," and 
a summary will suffice here. The figures state the relative humidity. 
Relative huutidity. * 
NOVKMBER 15-30. 
6 a. m. 12 nrwm. 4 p. m. 
Strand 88.8 70.9 83.3 
Interior 91.5 81.2 87.6 
DECKMBER. 
7.30 a. m. 11,30 a. m. 4 p. m. 
Strand 84.27 70.2 S0.r,2 
Interior 85.0U 79.28 . 81.85 
■ 
JANUARY. 
Strand 84.74 75.(i5 80.28 
Interior 84.13 74.05 80.00 
*A. Gray: Bot. iri?/.r.s l]xj>lor. Exped. (1854), 554. Caldeia, near wliich this 
tree was found, is 4 niih-s from San Ramon, and the collection was dou])th'ss 
from these groves. Maiden has confirmed the generic deturmination. Proc. V. /S'., 
Nat, Miis., Wash. (1903), 26, 091. 
"This Journal (190G), 1: 140. 
^ 
