Cohesion matters only because gravity gives up first. Shrink a body and its self-gravity falls away quickly, while the van der Waals pull between two touching grains does not care how big the parent object is, so below a certain size cohesion becomes the dominant thing holding the pile together, added to gravity rather than substituted for it [8]. That is what makes the spin barrier such a useful observation: asteroids larger than about 150 meters respect the 2.4-hour limit, which is what a gravity-bound rubble pile should do, and smaller ones spin faster than it allows [6][7].
The geometry of the test explains what the number means. Two boulders, each a meter across, with a matrix of fine cohesive grains packed between them into a bridge, drawn apart until the bridge ruptures [12]. With grains between 2 and 5 centimeters, each boulder is 20 to 50 grain diameters wide [19], which is roughly the ratio at which fine material behaves like cement between coarse blocks rather than like a continuous solid. It is the smallest system in which the effect of grain shape can be isolated, and Sanchez traces the idea back to Michael Swift of Nottingham and the liquid bridges that water forms between sand grains [5].
The sampling is thin per shape. Spread 78 runs across eleven shape classes, the sphere plus ten polyhedral forms, and it averages about seven runs each [20], before you also split them by uniform and mixed size batches [14]. A claim of universality therefore rests on those runs collapsing onto a single curve when plotted against sphericity and grain size, not on any one shape class being densely characterized [1].
The coffee comparison leaves out the number that matters. phys.org's headline puts Bennu's surface at 50 times weaker than ground coffee [18], which is a fine intuition pump and no substitute for a stress in pascals with the bridge geometry attached. The account also stops short of carrying the result from a two-boulder bridge to a whole asteroid, and it does not address how Bennu would answer a lander's push or a deflection impulse. The advance worth holding onto is narrower and more durable: cohesion between fine grains was a free parameter in this line of work from the 2010 proposal onward [9], the 2014 spherical-grain test had to assume it [10], and the material returned by OSIRIS-REx in 2023 let it be measured instead [11]. That is the real distinction: a strength law built on a measured input can be checked against reality, while one built on a fitted input was only ever built to match it.
