The violin is a highly complex vibrating system, which quite without the aid of science, evolved to a high level of sophistication. Wood, which varies considerably from one sample to another, requires individual attention to be fashioned into the plates of a good violin. It is not therefore surprising that mass-produced instruments are of very poor quality. It is the improvement of these instruments which is the objective of this thesis. After identifying those features of the response upon which the violin's quality most, depend a model is developed and used to answer several questions about violin design. As it is extremely difficult to include the sound post in the model it is suggested that an additional structural element, designed to match the back's impedance, is used to support the post. The amount of. acoustic radiation from the back is shown to be small so that this change does not greatly affect the output level. Finally it is shown that, using such a construction, the low frequency response of the violin may be predicted before assembly. A description of an automated production process in which the violin plates are cut and tested by micro-processor controlled machinery concludes the work.