The absorption spectra of small silver and gold clusters are presented. The experimental technique consists in depositing clusters in an argon matrix which is illuminated with white light. By collecting the outcoming light from the matrix and by analysing it spectrally, the absorption spectrum of the supported clusters can be obtained. To achieve these studies, a new experimental setup has been designed. This setup offers an enhanced sensibility and allows for the measurement of absorption spectra of systems that haven't yet been studied, like silver clusters made of 4, 6, 10, 12 and 14 atoms, as well as gold clusters. The two theoretical approaches used to describe the optical response of a cluster are briefly presented. The first one is the classical Mie theory applied to the Clemenger-Nilsson model, the second one being the quantum chemistry approach. The most relevant known results on silver and gold clusters are reviewed. The experimental setup is then precisely presented. The clusters provided by a sputtering source are mass selected and guided by sets of einzel lenses to the sample-holder, which is cooled down with a liquid helium cycle to approximately 28 K. The clusters are then co-deposited with argon atoms to form a matrix supporting clusters. The design of the sample-holder is presented along with its requirements and constraints. By using two different light sources, the absorption measurements are obtained over a large spectra, ranging from 2 eV to 6 eV. The method used to process the signal and yield an absorption spectrum is presented and discussed. The choice of a rare gas matrix is motivated by its weak interaction with the clusters. However, its effects on the optical response of the clusters are still important. These effects are reviewed and discussed in order to allow for a good understanding of the results that were obtained. Results found for clusters of Agn (n = 4, 6, 10, 11, 12, 14) as well as for clusters of Aun (n = 1 - 7) are presented and compared to existing measurements ans calculations. A good general agreement is found between our measurements and known results. In particular, the question of the coexsitence in the matrix of several geometries that are quasi-degenerate is asked. Finally, the validity of the Mie classical model to describe the shift of the spectrum due to the matrix environnement is discussed. The important relativistic effects in gold clusters are discussed through a review of theoretical calculations of the geometrical structures of these systems as well as through a comparison with results found for silver clusters.