The mass loss of C-rich giants

The mass loss rates, expansion velocities and dust-to-gas density ratios from millimetric observations of 119 carbon-rich giants are compared, as functions of stellar parameters, to the predictions of recent hydrodynamical models. Distances and luminosities previously estimated from HIPPARCOS data, masses from pulsations and C/O abundance ratios from spectroscopy, and effective temperatures from a new homogeneous scale, are used. Predicted and observed mass loss rates agree fairly well, as functions of effective temperature. The signature of the mass range M less than or equal to 4 M-circle dot of most carbon-rich AGB stars is seen as a flat portion in the diagram of mass loss rate vs. effective temperature. It is flanked by two regions of mass loss rates increasing with decreasing effective temperature at nearly constant stellar mass. Four stars with detached shells, i.e. episodic strong mass loss, and five cool infrared carbon-rich stars with optically-thick dust shells, have mass loss rates much larger than predicted values. The latter (including CW Leo) could be stars of smaller masses (M similar or equal to 1.5-2.5 M-circle dot) while M similar or equal to 4 M-circle dot is indicated for most of the coolest objects. Among the carbon stars with detached shells, R Scl returned to a predicted level ( 16 times lower) according to recent measurements of the central source. The observed expansion velocities are in agreement with the predicted velocities at infinity in a diagram of velocities vs. effective temperature, provided the carbon to oxygen abundance ratio is 1 less than or equal to epsilon(C)/epsilon(O) less than or equal to 2, i.e. the range deduced from spectra and model atmospheres of those cool variables. Five stars with detached shells display expansion velocities about twice that predicted at their effective temperature. Miras and non-Miras do populate the same locus in both diagrams at the present accuracy. The predicted dust-to-gas density ratios are however about 2.2 times smaller than the values estimated from observations. Recent drift models can contribute to minimize the discrepancy since they include more dust. Simple approximate formulae are proposed.