1. a square transparent cell: a square hole (11 x 11 mm 2 ) in a gold- plated copper plate (here 10 mm thick) 2 glass windows sealed with indium O-rings. this cell stands pressures up to at least 64 bar and can be cooled down to 35 mK. see Sasaki, Caupin and Balibar, submitted to J. Low Temp. Phys., July 2008. helium dendrites are 4 He crystals also facetted at high T ? S. Balibar, F. Caupin and S. Sasaki a Laboratoire de Physique Statistique de l'ENS (Paris, France) a - now at North Western Univ. (USA) conclusions and perspectives: It would be interesting to confirm this observation by a systematic study as a function of temperature and growth velocity. If confirmed, one possible interpretation of this re-entrant facetting could be that the liquid-solid interfacial energy strongly increases along the melting curve because the pressure P m (T) increases. Abstract: We have observed helium 4 crystals growing from normal liquid helium 4 at 2.58K. Their growth shape looks facetted, at least in the c, i.e. [0001] direction. This is a surprising observation since, up to now, the c facets were believed to disappear above the roughening transition temperature T R1 = 1.30 K. Do facets re-appear above 2K ? crystallization from the normal liquid mixing chamber pressure gauge fill line (0.6 mm ID) fast crystallization from the normal liquid at 1.87K produces a tangle of dendrites. Here this is obtained by opening a valve which connects the cell to a reservoir at 50 bar. The tangle scatters lights so much that the solid appears dark. Similar observations were done by Maekawa et al. (Phys. Rev. B 65, 144525, 2002) and by Ford et al. (J. Low Temp. Phys. 148, 653, 2007). In quench-frozen samples, Rittner and Reppy (PRL 2006, 2007, 2008) found «supersolid fractions» of order 20%. This large value might be due to such a large disorder At 2.58K, we were suprised to observe that helium snow flakes look facetted. Crystallization is obtained by injecting mass through the fill line at a moderate rate. Snow flakes are produced in the upper left corner where the fill line is connected to the cell. Their size is about 0.5 mm, homogeneous, close to the diameter of the fill line orifice (0.6 mm). They fall down in the liquid and often stick to the glass windows. They exhibit the 6-fold symmetry of the hcp structure. Those landing on the cell floor show that they are thin and flat. It seems that, at least in the « c » direction (ie the [0001] direction) the hcp crystals are facetted. This is surprising because the roughening temperature T R1 of hcp helium 4 crystals is at 1.30K (for a review on the surface of helium crystals, especially the roughening transition, see S. Balibar, H. Alles and A. Ya. Parshin, Rev. Mod. Phys. 77, 317, 2005.) If confirmed, and since the Kosterlitz-Thouless universal relation imposes k B T R =  (T R ) d 2 k B T R =  (T R ) d 2 where  (T R ) is the surface stiffness of the liquid-solid interface at the roughening temperature T R, it would mean that the roughening transition is re-entrant along the melting curve because the surface stiffness increases as a function of pressure, as predicted by H.J. Maris and F. Caupin (JLTP 131, 145, 2003). V.M. Giordano and F. Datchi (PRL 99, 165701, 2007) have found a re-entrant roughening in CO 2. It is also possible that replacing the superfluid by the normal liquid increases the interfacial energy, consequently the stiffness . facetted helium snowflakes facetted helium snowflakes T = 2.58 K P = 62 bar flat flakes on the cell floor helium snow flake falling down orifice of the fill line glass plate 2 crosses carved on the window to help focusing the camera helium snow flakes sticking to the front window T = 1.87 K 11 mm