In a previous paper one of the authors, A. Naruse, formulated the relation between the dimensions of the throat of a glass tank furnace and the maximum velocity of the current passing through it. And the present one contains the result of a series of model experiments carried out in order to built up a sound experimental background for the theory postulated before, and if necessary, to modify it.A transparent plastic model throat, 6cm long, 5cm wide, and variable height 1-4cm, having the dimensions as a whole which satisfy the similarity conditions to actual tank as far as possible, was used for the experiments. The model was immersed in a water bath which was composed of three compartments so that the temperature of either one of both end-sections may be fixed at a desired point. Furthermore, in order to set up the necessary temperatures at the principal points, the melting chamber was heated electrically with a hot plate, while the working chamber was cooled with water.As a liquid medium was used 83% glycerin, and as the traces for observing the flow of liquid were used the coloured oil droplets having the specific gravity well matched with that of the medium. At regular time intervals a fine line of coloured oil was drawn through the liquid, which immediately afterwards disperse into minute droplets. The velocity of flow at every level was determined from the photographic records taken at regular time intervals. The observations were carried out under the conditions both with and wothout pull.A fairly good agreement of the experimental values with those calculated by Peyches' formula was proved. And, if the correction for the friction at the side walls was introduced, the agreement became excellent.As long as the throat is not too high the velocity distribution in existence of pull was also in good agreement with the theory.The co-existence of the convection and the pull currents in a throat makes the forward current stronger, while it makes the backward current to run down gradually until it finally vanishes.This limited velocity Wc may be expressed asWc=1/72⋅ρ0ρbgk/η⋅Δθ/l⋅fh4When the height of the throat is changed keeping the pull velocity constant the limiting height having the same meaning as above may be represented as follows:f1/4hc=2.91(Wηl/ρ0ρbgkΔθ)1/4The symbols used in those equations are referred to the table of nomenclatures in the text.From the technological point of view the authors maintain their opinion that both figures give the corresponding optimum values which reduce the velocity of erosion of refractories to the minimum.