Inspired by the familiar phenomenon of the cherry entanglement, we com
pute the clustering properties of a new model in which pairs of elemen
ts,located on the sites of a lattice and tiling it completely, attach
to form aggregates of various sizes. Avoiding double-occupancy, the di
stance between the two elements of each pair is chosen at random with
a prescribed maximum length, l. The resulting clustering behaviour cru
cially depends on the dimension of the embedding lattice and on l. We
devise a transfer-matrix method to solve the model in 1 dimension. Res
ults in 2 dimensions are obtained by Monte Carlo simulations. In 1 dim
ension, we define a characteristic cluster size: it depends exponentia
lly on l. In 2 dimensions, for a big enough 1, we find scaling for the
probability of occurrence of clusters of size n versus n. Possible ge
neralizations of this model are qualitatively discussed.