We present recent results from a program devoted to the study of small scale structure in translucent molecular clouds, using dust as a tracer. Several methods have been employed: i) statistical analysis of stellar fields, ii) studies of background galaxies; iii) searches for time variations of the extinction and reddening. For each method, we summarize the principles, the type of constraints provided (scales, sensitivity) and present the results obtained so far. We conclude by some prospects concerning direct studies of the distribution of H2 itself.

Motivations

Originally, we wished to constrain in a direct way the level at which the penetration of the stellar UV and visible radiation is enhanced by structure effects (cf Boissé 1990). Studies of spatial variations of dust extinction and reddening offer a powerful way to address this question. Indeed, in the presence of density fluctuations, the analysis of extinction in stellar fields directly provides a measure of the effective opacity (Thoraval et al. 1997). Further, in the visible, the required observations are easy to perform and provide an excellent spatial resolution.

Another motivation comes from the detection of very small scale structure both in atomic (Frail et al. 1994) and molecular gas (Moore & Marscher 1995). In the assumption of a uniform dust to gas ratio, one should observe corresponding variations of the amount of dust, resulting in local variations of the extinction.