8 PROCEEDINGS OF THE NATIONAL MUSEUM. VOL. 62. 
silicate glass was used) crystallizes as quartz below 870°. Although 
quartz is the stable form below this temperature, either amorphous 
silica or cristobalite changes first to tridymite and then to quartz only 
after much longer heating. Between 870° and 1,470° tridymite is al- 
ways formed in a tungstate melt. Within the tridymite range amor- 
phous silica heated in fused sodium tungstate or alkali silicate yields at 
first a mixture of cristobalite and tridymite, which later becomes 
entirely converted to tridymite. From 1,475° upward any form of 
silica heated in a tungstate melt is change to cristobalite. Heated 
without a flux quartz is changed to cristobalite, while quartz and 
tridymite are never formed in the absence of a flux. In addition to 
these inversions, which are marked by a complete change of crystal- 
line form and which take place slowly and with difficulty, there is 
another class of inversions of different character which take place 
with no notable change in the outer form of the mineral. Cristobal- 
ite is isometric in form but is always feebly birefringent and uniax- 
ial optically. The isometric form is stable only at an elevated tem- 
perature and inverts upon cooling to the double-refracting pseudoiso- 
metric form. These are known as a- and f-cristobalite, respectively. 
Thus an ordinary birefracting crystal of a-cristobalite heated to a 
moderate temperature, 300° or slightly less, becomes converted to 
more transparent isotropic #-cristobalite, which on cooling reverts 
again to the birefracting a-cristobalite. The variability of the tem- 
perature at which this inversion takes place has been discussed in the 
paper cited and need not be gone into here. The optical and physi- 
cal properties conclusively isolate cristobalite from the other modifi- 
cations of the compound SiO,. 
In those specimens of light-colored nonopalized rock from Spokane 
which show plagioclase crusts or coatings of the gray enamellike 
feldspar there occur, associated with the octahedrons of magnetite 
and attached to the plagioclase or the gray glaze, small whitish 
crystals or rounded bodies of a porcellanous white mineral agreeing 
in appearance with the cristobalites described by Rogers. When 
boiled in hydrochloric acid, these show no tendency to dissolve. 
This mineral, which had provisionally been called analcite by Mr. 
Fair, was examined optically by Mr. A. Rodolfo Martinez in this 
laboratory and found to agree with cristobalite in optical properties, 
its mean refractive index being 1.485 +0.003. The mineral occurs in 
whitish crystals averaging 0.5 mm. in diameter, which under a binoc- 
ular microscope were seen to vary from simple forms to almost spheru- 
litic bodies. In some ways the least simple of these resemble complex 
penetration twins which could not be interpreted. Occasional 
crystals, though dull, have the perfect cuboctahedral form shown 
in figure 6. Most of those which are of this form are also imperfect, 
and one which was mounted on the goniometer gave no dependable 
