Heat transfer and fluid flow characteristics for natural convection of air in a vertical parallel channel system have been examined. The results of both theoretical and experimental work are reported in this thesis. The system of differential equations governing the fluid flow was solved using the PHOENICS code program. The PHOENICS solution gave velocities, temperatures and pressures throughout the field and velocity and temperature profiles are presented at different vertical locations for three channel heights, 1,2 and 3 m, each for three different channel widths, 50,100 and 150 mm. A further PHOENICS sub-routine was written to obtain relevant dimensionless parameters such as the Nusselt Number, which characterises the heat transfer to the air, the Rayleigh Number, the dimensionless air flow rate and dimensionless channel length. The results obtained have been compared with experimental results and with existing data for channels with constant wall temperatures. Non-isothermal wall cases were also considered and the resulting velocity and temperature profiles are presented. The channel used in the experimental work was formed from two vertical plates, one of which was electrically heated, while the other was glass. Heated plates of height 1m and 2m were used and combination of these formed a plate of height 3 m. The width of the plates was 1 m. A double glazed cover plate of the same dimensions could be adjusted to give spacings between the heated plate and glass from 25 mm to 150 mm. The electrically heated plate could be controlled to give the required constant plate temperature (range 35-125°C) and heating could be augmented by a solar simulator consisting of 50 Tungsten Halogen Lamps of 150 Watts each. From the experimental results, relationships between the Nusselt number, Rayleigh number, dimensionless air flow rate and dimensionless channel height have been obtained. In addition, the effect of diffuser sections at the channel inlet and outlet and transient operating conditions were investigated experimentally. Effects of atmospheric pressure and humidity were also considered. The experimental results are compared with those from the PHOENICS solution and with existing data for constant wall temperature conditions and they show good agreement. A discussion of the use of the correlating equations for the heat transfer coefficient and air flow rate in the design of passive solar heating systems, such as the Trombe wall, is also included.