Large prominent families contain several hundred recognized asteroids (and many more smaller objects which may be either not-yet-analyzed, or not-yet-discovered).
Such cratering families typically consist of a single large body and a swarm of asteroids that are much smaller.
Such small asteroids then become subject to perturbations such as the Yarkovsky effect that can push them towards orbital resonances with Jupiter over time.
Further evidence for a large number of past families (now dispersed) comes from analysis of chemical ratios in iron meteorites.
These show that there must have once been at least 50 to 100 parent bodies large enough to be differentiated, that have since been shattered to expose their cores and produce the actual meteorites (Kelley & Gaffey 2000).
The Japanese astronomer Kiyotsugu Hirayama (1874–1943) pioneered the estimation of proper elements for asteroids, and first identified several of the most prominent families in 1918.
The most prominent algorithms have been the hierarchical clustering method (HCM), which looks for groupings with small nearest-neighbour distances in orbital element space, and wavelet analysis, which builds a density-of-asteroids map in orbital element space, and looks for density peaks.
The boundaries of the families are somewhat vague because at the edges they blend into the background density of asteroids in the main belt.
Additionally, some interlopers from the heterogeneous background asteroid population are expected even in the central regions of a family.
Other terminology has been used to distinguish various types of groups which are less distinct or less statistically certain from the most prominent "nominal families" (or clusters).
Tribes are groups that are less certain to be statistically significant against the background either because of small density or large uncertainty in the orbital parameters of the members.