946 THE AMERICAN NATURALIST. [VOL. XXXV. 
The first part of Vol. V of the Flora Capensis, edited * by 
Sir W. T. Thiselton-Dyer, extending from the Acanthacez to Clero- 
dendron in the Verbenacez, has recently appeared from the press 
of Lovell Reeve & Co., of London, — unfortunately without indica- 
tion of the date of issue. 
An interesting essay on palms and their value to residents of 
tropical countries, by Professor Schróter of Zürich, constitutes the 
last /VeujaArsb/att of the Naturforschende Gesellschaft of that city. 
In connection with a monograph of garden beans recently pub- 
lished from the Missouri Botanical Garden, a paper by Professor 
Halsted on bean diseases and their remedies, constituting Bulletin 257 
of the New Jersey Agricultural Experiment Station, becomes of double 
interest. Like all of Professor Halsted's papers, it is well done. 
PETROGRAPHY. 
Origin of Corundum. — The interesting problem relative to the 
origin of corundum in basic rocks has been attacked by Pratt' 
through the study of slags. He finds that the separation of corun- 
dum from magmas is dependent upon the composition of the magma, 
upon the character of the oxides dissolved in it, and upon the 
quantity of alumina present. 
When the magma is a calcium-sodium silicate, corundum separates 
only when the ratio of Al O; to the other bases exceeds 1: 1, and the 
ratio of SiO, is less than 6. If Mg and Fe are present, no corundum 
will form unless there is more than enough ALO, present to unite 
with these bases. 
When the magma is a magnesium silicate without excess of Mg, 
all the AlO; will separate. If Mg is in excess, some of the Al;O; 
will unite with this, forming spinel, and the remainder will separate. 
When Cr;O, is present and only a little AlO; and MgO, these unite 
with the Cr,O;, yielding chromite. No corundum is formed. 
When alkalies or alkaline earths are present, the Al,0; tends to 
unite with these in the formation of feldspars. There is, however, 
little tendency to the formation of Mg-Al silicates when the magma 
is a magnesium silicate. 
1 Amer. Journ. Sci. (1899), p. 277. 
