Does niche limitation exist?

1 Three sets of data are examined for evidence of niche limitation expressed as a deficit of variance in the number of species per sample, as compared to a null model with species occurring independently of each other. 2 An agricultural field of permanent pasture was sampled with contiguous quadrats. It showed no variance deficit but rather a significant variance excess. Subdividing the field floristically or environmentally to achieve greater homogeneity did not reduce this excess. 3 A fixed dune slope was sampled by a line of contiguous quadrats and analysed as quadrat sizes in the range 20 mm X 20 mm to 1 m X 20 mm. There was no variance deficit at any quadrat size; at quadrat sizes up to 100mm X 20mm there was significant variance excess. 4 A reanalysis of Patrick's (1968) diatom data gave a variance deficit one year (but P = 0 5) and an excess the next. 5 It is questionable whether the concepts of niche limitation, and hence of species packing, are meaningful if they cannot be demonstrated even in favourable circumstances.