Rotational symmetries of macromolecules are most clearly perceived in
the en face projection and may be assessed by inspection of rotational
power spectra calculated from electron micrographs of individual part
icles. However, if the symmetry is not contrasted strongly, this proce
dure may be inconclusive since the relevant peak may not be convincing
ly higher than other spectral components. To some extent, this is a sa
mpling problem since the number of repeating elements involved is usua
lly small. We have devised more sensitive statistical tests for rotati
onal symmetry that pool the information contents of entire populations
of particles. Both tests involve combining the rotational spectra of
many particles and comparing them with the spectra of surrounding back
ground areas. One method is based on the well known t-test which estim
ates whether two populations differ at a given significance level. In
the second test, the ratio between the intensity of each component of
the rotational spectrum and the average corresponding intensity for ba
ckground areas is calculated, and thence, the cumulative product of th
ese ratios over all particles in the data set. If a symmetry is presen
t, this product gradually diverges; otherwise, it converges to zero. A
s a practical trial, the tests were applied to micrographs of negative
ly stained hexons of herpes simplex virus and confirmed their 6-fold s
ymmetry. Applied to negatively stained ''connector'' proteins of bacte
riophage T7 purified from a plasmid expression system, both algorithms
detected polymorphism with distinct subpopulations of both 13-fold an
d 12-fold connectors.