Since publication (1) the apparatus previously described by the authors has been altered and improved. The alterations have been concerned mainly with excluding draughts and securing a better form of support. In using the apparatus we have found that the absence of vibration is important for the agreement of duplicates. It would seem that vibration tends to break the cohesion between the curd and the surface of cylinder A. For this reason a part of the laboratory as free as possible from vibrational disturbances was chosen and the support fixed to the wall.