A brief survey is given of the ideas underlying the interpretation of molecular vibration spectra in connection with molecular structure problems. This is followed by a discussion of the reasons for the appearance or nonappearance in the spectrum of the characteristic frequency at 3400 cm—1 for various molecules containing the OH group. In certain alcohols and glycols this frequency has not been reported in the Raman effect, but it is concluded that this must be due to experimental difficulties. In the carboxylic acids, on the other hand, as well as in certain aromatic compounds containing OH, the characteristic frequency is definitely absent from both the Raman and infrared spectra, and the role of the hydrogen bond in association and chelation is discussed in connection with these cases.