Published diffraction data are critically reviewed, and replotted in a new way to show the variation with concentration of the 8- to 25- nm diffraction maximum. Most of the early data are found to be consistent with a single model for a liquid-type array of mutually repulsive particles, whose molecular weight is calculated to be that of a nucleosome or possibly a dimer. The data for all but the highest concentrations, where distortion due to dehydration is possible, support no particular model for the higher-order coiling of chains of nucleosomes, and cannot be used to support models for "native" chromatin. Only in the presence of excess salts or after isolation with polyamines is there aggregation in solution of nucleosomes, which then give peaks at 11 and 5.5 nm that do not change much with concentration. Recent work by the authors confirms that under some conditions nucleosome undergo a transition to a state whose diffraction is consistent with hexagonal packing of extended DNA to which histones are still attached. This state is probably responsible for much of the strong 2.7-nm peak previously obtained from certain samples, which was in some cases assigned to nucleosome structure. Only the peak at 3.7 nm is clearly attributable to the form factor of the isolated native nucleosome.