Fig. 1: Fragment size distributions of stressed and unstressed materials.

The error bars in the graphs represent one standard deviation and are inferred by viewing the binning as a Bernoulli process. a Photograph of a Prince Rupert’s drop clenched by a plier (left), and the fragmentation of a (different) Prince Rupert’s drop triggered by cutting the tail with the plier (right). See Supplemental Information for a high-speed movie of the fragmentation of a droplet. b Tempered glass plate fractured by impact on one of its sides. For illustrative purposes the back of this particular plate is covered by adhesive tape so fragments are still fixed together after fragmentation, showing the crack pattern. c Fragmented unstressed glass plate, with the big fragments puzzled back together after fragmentation. The sizes that range from the size of the system all the way down to tiny fragments, illustrate well the power-law behavior in this case. d Fragment size distribution of the Prince Rupert’s drop shown in a, where d is the equivalent spherical diameter. One drop fragments in at least 21,847 fragments, measured by micro-CT. Two exponential regions can be identified, which indicates inhomogeneous fragmentation. The solid line is a fit of the form \(\,\text{p}\,(d) \sim {C}_{1}\exp (-d/{d}_{1})+{C}_{2}\exp (-d/{d}_{2})\), with d₁ = 0.31 mm and d₂ = 0.064 mm. The smallest fragments are what in most other experiments is referred to as the “remaining dust”. e Fragment size distribution of the tempered glass plates shown in b rescaled by d_c for each plate thickness, where d_c is the characteristic length set by the location of the maximum. The distribution is a truncated exponential distribution (as indicated by the solid line) with an exponential cutoff at a characteristic length set by d_c. The dotted line is a generalized logistic distribution, fitting the full distribution well. f Fragment size distribution of two unstressed glass plates, where the solid line is a power-law fit with an exponent of 1.9. Source data are provided as a Source Data file.