Journal article Open Access
A large silicate glass density database including 6719 compositions, obtained from SciGlass® and other published sources was analyzed statistically through linear regression for evaluating the global agreement among room temperature density data reported in the literature in 1030 different studies from various investigators, and for facilitating efficient estimation of the density based on a given composition. It was found that the literature data agree very well. Using just 101 of the initial 6719 degrees of freedom 99.8% of the measured density data variance could be reproduced. The standard error of the developed model was 0.019 g/cm³.