ABSTRACTIt is unjustifiable to assume that a furnace roof cannot be safely insulated. The principal variables controlling the possibility of safe insulation are: (1) temperature of bath below melting point of silica brick roof; (2) transparency of the atmosphere within the furnace,i.e., no possibility of impingement of dense hot flames. The data of this paper are taken from observations on a glass tank furnace in which the combustion rate (B.t.u. per cu. ft. per hr.) is very low. A testing method is described which will show whether or not insulation is safe for any given furnace, considering only the temperature of the inside face of the brick as affected by the application of insulation. The possibility that if the roof is insulated the progress of fluxing of the silica brick by dust and fumes within the furnace may be accelerated because the higher mean temperature of the brick is considered. In the furnace observed, corrosive dust and fumes were in the gases to a considerable extent, yet no unusual deterioration of the roof brick was noted after three years of insulation under continuous operation.