To the Editor:

Villermaux and Bossa1 (hereafter referred to as VB) presented a model to describe bag break-up of large drops (diameter d ≥ 6 mm), which yields a drop size distribution consistent with Marshall–Palmer's distribution for rainfall2. Based on this result, and a laboratory demonstration of bag break-up, VB propose that the equilibrium drop size distribution of natural rainfall is the result of the spontaneous fragmentation of single large drops (d ≥ 6 mm), and that the contribution of drop–drop interactions such as coalescence and collisional break-up is negligible. This proposition is ill-founded. First, the equilibrium Marshall–Palmer distribution is not representative of the variability and transient nature of natural rainfall3. That the VB break-up model yields the Marshall–Palmer distribution is therefore not sufficient to establish that it is physically relevant. Second, field observations show that the number concentration of large drops (d ≥ 6 mm) if observed is extremely small4,5. VB do not address the physics required to explain the population dynamics of such large drops consistent with the duration, intensity and microstructure of natural rainfall6. Third, the laboratory experiments reported by VB do not replicate the free-fall conditions of natural rainfall7,8.

Whereas evidence of bag break-up is limited9, collisional disk break-up that results from the fragmentation of a large drop originated by the near head-on collision of two drops is well documented10. The transient shape of the large drop evolves to form a 'bag' after collision, and then a 'disk' before breaking to yield a large number of small drops. VB speculate that collisional break-up is unlikely. However, the mean free path for drop–drop collision is of the same order of magnitude as the average distance a drop has to fall before experiencing spontaneous break-up3. Furthermore, spontaneous break-up occurs only once for each (rare) large drop, whereas many interactions can occur among a wide size range of small drops in their downward trajectory. Indeed, surveys of laboratory experiments under free-fall conditions7,8,9 indicate that collisional disk break-up is responsible for up to 20% of all break-up events7,8 (other types of break-up being filament and sheet break-up7,8,10), whereas none or singular observations of spontaneous break-up are reported. VB's thesis that single-drop fragmentation determines the size distribution of raindrops is in conflict with field and laboratory evidence reported so far.