This report describes the detailed measurements in an experimental investigation of flames spreading under an incombustible ceiling. Later reports will deal with combustible ceiling linings and the practical applications of the work. Experiments were performed in a model representing the ceiling of a corridor with a fire at one end - a gas burner was used to represent the fire, this was replaced by wooden cribs in experiments to be described in a later report. Depending on the size of the fire and its distance beneath the ceiling, the flames may have drawn up into the horizontal layer of hot gases enough air for combustion of the fuel gases. The flame length is then determined by mixing within the layer. The air drawn up may not be sufficient for the combustion of the gases from large fires and the remaining air required for combustion is entrained vertically into the hot gases flowing over the cooler air beneath. Correlations of horizontal flame lengths have been derived and related to the much shorter lengths of vertical flames. Relationships have been derived from the experimental data from which it is possible to estimate the radiation downwards from the hot ceiling and the gases beneath it to the floor with a view to estimating the contribution to fire spread on the floor. A heat balance of the ceiling gases was satisfactory, so confirming the validity of the calculations. Horizontal flames radiate more of the heat produced at a level sufficient to assist fire spread than do vertical ones.