Previous perturbation treatments of the Amide I vibrations of 8 polypeptides are inconsistent with a detailed normal coordinate analysis of crystalline polyglycine I. This analysis indicates that the Do interaction constant is essentially zero, rather than the large value (about 20 cm-') required by the earlier application of the perturbation theory. It is suggested that the previously neglected Di, term should be included in the perturbation expression, and it is shown that the physical origin of such a term can be accounted for by transition dipole coupling. This mechanism is shown to give a reasonable explanation of splittings of the C=O stretching vibrations in hydrogenbonded carboxylic acid dimers. Its application to , polypeptides provides a satisfactory interpretation of splittings in the Amide I modes.The useof vibrational spectra to identify different polypeptidechain conformations dates from the work of Elliott and Ambrose (1), who observed that the frequency of the infrared-active Amide I vibration of a polypeptides was about 20 cm-' higher than that of # polypeptides. Further studies on synthetic polypeptides (2) established this observation as a firm empirical rule.Early attempts (3, 4) to account for this difference were not fruitful. A significant advance in understanding the theoretical basis for the above observation was achieved by Miyazawa (5), who developed a perturbation treatment for the interaction of amide-group vibrations in various polypeptide-chain structures. He also noted that, in addition to the strong Amide I band at about 1630 cm-' in the # polypeptides, a weak band at about 1690 cm-' could be assigned to the antiparallel-chain pleated-sheet structure. This treatment has formed the basis for interpretation of the vibrational spectra of polypeptides and proteins (6, 7).In the Miyazawa theory, the frequency of an Amide I mode (which is the only one that we will be concerned with in this paper) is given by p(B,V') = vo + E D,, cos(s8) cos(t') [1] In applications of Eq.[2] to the antiparallel-chain pleatedsheet structure, v(O,Tr) and v(r,O) were taken from the spectrum of polyglycine I, and the frequency of nylon 66 was The above procedure has been criticized (9, 10), primarily on the grounds of the assumed constancy of vo. This criticism was based on studies of a series of polyamides (10) that indicated that jo varied by 6 cm-I between nylon 10 and nylon 2. Such variations can give rise to errors in the interaction constants Do and Do, that are of the same order as the constants themselves. This objection to taking vo from nylon 66 as applicable to polypeptides has a general validity, since it should not be expected that the unperturbed frequency associated with the peptide group is completely independent of the local chemical environment of this group: This is amply established by detailed normal coordinate analyses of N-methyl acetamides and nylons (11), as well as of related molecules (12).It is also confirmed by an extension of these calculations to polyglycines I (13) ...