ANALYSES OF ROMAN GLASS FROM SILCHESTER. 165 
sideways. This glass contains less iron than the above, 
the amount of manganese is correspondingly small; the 
materials were probably specially selected.—(See analysis 
B.) 
Green glass.—Fragments were picked out showing a 
fairly deep bluish-green tint; they contained many blebs 
and consisted apparently of pieces of the necks, handles, 
and bottoms of bottles or vases.—(See analysis C.) 
From these three analyses it is impossible to say definitely 
whether manganese was purposely added or not, it probably 
was, but it may have been naturally present in the materials 
used. The question requires further investigation.—A. 
LIVERSIDGE. | 
Method of Procedure.—In the general analysis the 
methods of Washington’ and Hillebrand’ for a naturally 
occurring silicate were adopted in the main. The more 
important features may be briefly indicated. 
Specific Gravity.—The specific gravity of two fragments 
of Specimen A, was found by a Jolly balance, the results 
being 2°49 and 2°50 respectively. Check determinations 
by the ordinary balance gave 2°493 and 2°499. 
Total Water.—For this determination Penfield’s’ method 
was employed. It consists essentially in igniting the 
powdered substance in a narrow glass tube closed at one 
end—condensing the moisture on the cooler part of the 
tube—drawing off this portion and weighing it, with and 
without the moisture. The results obtained were very 
consistent (the greatest difference being °02% in a series of 
four) and the many possible sources of error of the old ‘loss 
on ignition ’ method seem to be avoided. 
* Washington—Manual of Chem. Anal. of Rocks, 1906. 
? Hillebrand—‘Some Principles and Methods of Rock Analyses,” Bull. 
U.S. Geolog. Sur., 176, 1900. 
* S. L. Penfield—Amer. Journ. Sci., xivirr., p. 31, 1894. 
