It is suggested that the distribution of DNA molecular lesions (i.e.,
molecular lesion spectrum, MLS) produced by a particular electromagnet
ic radiation characterizes that radiation, and that these distribution
s may serve as radiation signatures. These signatures can have the adv
antages of (1) high fidelity, (2) persistence times at least ten order
s of magnitude greater than track structures, and (3) a space-time cel
lular position which is approximately half way between initial insult
and ultimate biological dysfunction. Spectra are presented for 405nm,
254nm and gamma-ray radiations. These MLS's are visually distinctive.
Comparable data bases for most other radiations, particulate and elect
romagnetic. are generally incomplete. We propose the application of ad
aptive feature recognition and classification techniques to the extrac
tion of radiation signatures for such incomplete data sets. In this wo
rk, however, such an application is restricted to quantitation of the
degree of orthogonality of the three patterns (i.e., MLS's) for 405nm,
254nm and gamma-rays.