Simple statistics shows that heat-treated glass is much stronger than expected.

In structural design the bending strength of pre-stressed glass (HTGS) - heat-strengthened or tempered - is usually formulated as the sum, weighted with correction coefficients, of the values for pristine-material strength (PMS) and surface pre-compression (SPC), induced by the thermal treatment. Characteristic values are associated with the 5% fractiles of the corresponding statistical distributions, but since they are different for PMS and SPC, the 5% fractile for the distribution of HTGS may be much higher, due to statistical interference, than the simple sum. Assuming for PMS a two-parameter Weibull distribution and for SPC a Gaussian distribution, as suggested by experiments, the probability density function for HTGS is obtained by statistical convolution. Several aspects, such as test set-up, can affect the difference between the sum of the fractiles of the operant distributions and the fractile of the compound distribution. This finding suggests that the formulas presented in most standards for HTG strength may be too much conservative. Of course, an ad hoc experimental campaign is necessary to corroborate this theoretical finding that, if confirmed, could permit a much better use of the material, with incommensurable savings in construction works.