An additive formula is given for the computation of specific volumes of molten and compacted alkali silicate binary glasses with modifiers up to 50-mole percent and from annealing temperatures (400 0 or 500 0 C) to 1,400 0 C. The effective partial volumes, Vs, for the silica. are postUlated as (l /vs ) = 2.198+r.C8 (1723 -t), where r .. is the mole fraction of nonsilica and Cs is a constant to be evaluated from glassdensity data. This is based on the idea that silica networks can contract in volume as temperatures are lowered provided, and in proportion as, modifier ions are present in the glass. The effective partial volumes of R 20, the non silica, are assumed as linear function s of temperature but not of the fraction of silica present.Computed and observed densities agree within approximately 1 p ercent. The effective volumes of nonsilica are somewhat smaller than published estimates for the oxides themselves, as should be expected because of interpenetrations.Previous formulas for the computation of densities of glasses have related principally to variations in composition. The formula proposed by Stevels [1] 1 is based on ideas of structure and has very few constants. It assumes for restricted ranges in composition that volume is independent of the nature of the cation. As modified by Stanworth [2] it shows that volume is dependent on the size of the cations, the glass volumes being larger as cation radii are larger.The additive formula proposed by Huggins [3] has many constants, and it is notably successful. That work is of considerable interest to anyone interested in structure because Huggins finds that the addition of every cation oxide molecule linearly increases the volume of a glass per oxygen atom, regardless of previous additions. However, at least for R 20 silicate glasses, it can be shown that the linearity is not dependent on the referral to oxygen, and his corresponding partial volumes of nonsilica (not referred to oxygen) also are found to be linear if plotted against mole fraction of the added oxide. 2 The Huggins formula is of further interest because the partial volumes used for silica are almost identical in the presence of various kinds of nonsilicas and are so treated except for the presence of a few elements, notably boron. Most interesting of all in connection with the present paper, his partial molar volumes (P. M. V.) attributable to silica 3 decrease as larger amounts of modifier are added. The Huggins stepwise procedure in dealing with the silica component is roughly equivalent to the procedure mentioned by LeChatelier [13], namely an arbitrary dual 1 Figures in brackets indicate tbe literature references at tbe end oC tbis paper.• Altbough tbe Huggins partial volumes Cor silica (wbetber or not referred to oxygen) are more nearly representable by a smootb curve wben plotted against mole traction oC silica tban against tbe Si to 0 ratio (used by Huggins), tbeir better representation by segments of straigbt lines can still be shown. The Huggins special points of intersection corres...