Glass containers are relatively heavy. Also, glass is fragile, and breakage sometimes occurs during handling and transport. Therefore, glass bottles must be constructed to achieve maximum strength at minimum weight (wall thickness). To our knowledge, about 85% of all breakage is caused by external impact. The purpose of the present paper is to study the relationship between the shape and thickness of glass bottles and the impact loads they can resist by using linear multivariate statistical/mathematical regression (or calibration) techniques (UN-SCRAMBLER), in order to compute minimum required thickness of the bottle as a function of impact strength and vice versa. The study was based on 10 different types of bottles. Moreover, we have concentrated on measurements related to the heel of the bottle. The set of bottles used are described as follows: returnable, round body, straight side wall and without metal oxide coating. All the bottles were given a standard abusement before the impact tests. The following conclusions were reached: 1) there are strong relations between glass thickness and the resistance to external impact (as expected); 2) iimultivariate calibration gave much better results than using only one variable at a time; 3) the predictive ability is not good enough (accuracy of ±10–15%). But provides useful information that would be difficult to obtain by other methods.
