This dissertation is concerned with the physical oceanography of meltwater plumes emerging at the base of glaciers at heads of fjords. First, a numerical model is developed and simulations to show the use of such a model are presented. Field work to verify the model was carried out, and the results of these surveys are discussed in conjunction with other field data. Finally, laboratory experiments are presented, simulating the scenario of a plume of meltwater emerging at the surface of the fjord some distance from the glacier face. A brief summary of each of the three areas of research are given below. Circulation and mixing resulting from melt-driven convection in the proximity of ice have important oceanographic consequences, as well as being of interest to biologists studying organisms in these regions. A simple steady state, one-dimensional numerical model has been developed to simulate the characteristics of a cold, fresh meltwater plume emerging at the base of a vertical glacier face, at the head of a fjord. The importance of the initial width, speed and temperature of the outflow is shown. Simulations showing meltrates with depth are presented, and profiles of the retreat of the glacier face are given. Sensitivity of the model to varying the entrainment constant and the slope of the ice face are examined, as well as the sensitivity to other parameters of the model. Relevance to the Ice Pump Mechanism is made. Field work was undertaken in a fjord on the west coast of Spitsbergen, Svalbard to verify the results and characteristics of the meltwater plume of the numerical model developed. The data showed the fjord to agree well with previous work carried out in fjords with glaciers at their heads. Using a simple approach, the dynamic method, and from observations, it was conjectured that the Coriolis force has an effect on the circulation in the fjord. Data collected parallel to the glaciers in the fjord supported this conjecture. The field data used as comparisons comprised of a set from a fjord in the southern hemisphere, with a glacier at its head, the data being processed by myself, and a set from a fjord system north, along the coast of Spitsbergen, from where the author performed her own field work. These two comparison data sets were collected by others. In the series of laboratory experiments it was found that if a line plume is located near the closed end of a channel, a filled region of fluid forms near the end wall and a gravity current propagates towards the open end of the channel. In the experiments a steady state is attained in which the gravity current has approximately one half the depth of the channel and the filled region near the wall has the same density as the fluid in the gravity current. A simple theory is described to model the experiments and predict the speed and concentration of the gravity current. The situation is very different from the case where the plume is at the end wall, in which case no filling-box region develops and the gravity current remains significantly thinner and more concentrated. A physical model of the process is derived and tested with quantitative experimental results. Comments on the limitations and applications of the experiments and the theory are made. The final part of.the thesis discusses model, field and laboratory results. Reasons for direct comparisons between the three areas of work not being possible are discussed. Suggestions for future work concerning numerical modelling, field work and laboratory experiments, by which this research could be extended are made.