In calibrating any apparatus adapted to measuring directly the coefficient of absolute viscosity of liquids, the constant must be obtained within certain limits of error. Direct calculation would leada priorito about 1 per cent for this error. The experiments made up to date show that it is more and that it may be 2 per cent. The authors attribute these additional errors to the great variation of viscosity with temperature of the standard liquids used, and to the uncertainty in maintaining uniformity of temperature in these liquids.The custom of using only such viscometric apparatus that is capable of giving directly a correct measure of this physical constant is spreading more and more among technologists of the oil industry. More and more often the objection is raised that these instruments are rather delicate in manipulation. The one thing really necessary for transpiration instruments is to have a set of capillary tubes, which assures laminar streaming, the so‐called Poiseuille régime, during every determination.Since it is impossible to construct capillaries of predetermined dimensions, it is necessary to calibrate each tube before releasing it for service.It is in a course of preliminary experiments toward the establishment of a center for the calibration of viscometrical tubes at the National School for Petroleum at Strasbourg, that we have been led to make the above remarks. We should like to have the opinions of others who have had occasion to do similar work.