The freezing behavior of supercooled waterdrops nucleated by contact with clay particles was studied by means of an experimental setup recently described in detail by Pitter and Pruppacher (1973). This setup utilizies the UCLA cloud tunnel, which allows stable suspension of supercooled waterdrops and ice particles in a vertical airstream and also allows retrieval of the frozen particles from the airstream for further study under a microscope in a walk-in cold room. By using this setup the mode of fragmentation of freezing waterdrops was investigated. It was found that, depending on the drop surface temperature up to 42% of the drops frozen by contact nucleation with a clay aerosol transformed into ice particles with surface breaks that resulted from ejection of 1 splinter, whereas only up to 8.5% of the frozen drops shattered and split into 2-3 fragments, and only up to 6.3% of the frozen drops totally ruptured into a large (>3) but undetermined number of very small ice fragments. Drops frozen by clay particles submerged in a drop did not shatter and produced much fewer frozen drops with surface breaks. Our results are discussed in terms of the present knowledge on mechanisms that lead to ice multiplication in atmospheric clouds. It is concluded that, in agreement with earlier measurements reported in the literature, ice multiplication by freezing and splintering of freely falling drops indeed cannot account for the large concentration of ice crystals found in certain clouds with cloud top temperatures warmer than -10¿C. On the other hand, the efficiency of this mechanism is not low enough to disregard it altogether. In particular, it may contribute to getting the ice multiplication mechanism started, which, according to Hallett and Mossop (1974), is efficiently carried on by the production of ice splinters during riming, once ice particles sufficiently large for riming are present in a cloud.