Subthreshold events (pion production, for instance) at energies E < m π are treated as rare events. The associated multiplicity distribution P c (n) and the unconstrained distribution P (n), when there is no rare event-trigger, are related by the model independent relation P c (n) = n 2 n 2 P (n) .This relation, and in particular its improved version inspired in clustering of nucleons, is in fair agreement with data. Moreover, it allows to extract from the multiplicity data information on the number of nucleons involved in meson production processes, occuring at energies below the production threshold in free nucleon collisions.1 Production of particles, pions or other hadrons, energetic photons or fast protons, in nucleus-nucleus collisions at energies per nucleon well below the free nucleon-nucleon threshold for such production [1] gives clear evidence for nucleonic correlations in nuclear matter. These correlations may simply reflect the fermionic nature of the nucleons or they may be seen as true collective effects, like clustering of nucleons (see [2] for a review).By clustering processes we mean here pion production mechanisms involving sub-systems with a number of nucleons between 2 and the mass number A, the simplest example being the interaction of a nucleon with a deuteron. In fact, schematically we can represent pion production in the pd→ π 0 pd reaction via resonance formation as in the diagram of Fig.1a. The threshold for the mentioned reaction is lower than the free nucleon threshold (Fig.1b). In contrast to this diagram, the one of Fig.1a is kinematically allowed at subthreshold production energies, because the two nucleons of the deuteron are correlated. Basic diagrams [3][4][5][6], like the one in Fig.1b, require a medium to be non vanishing, for energies below the production energy threshold.Higher mass resonances can be obtained if larger clusters supply the required energy. Our emphasis here is on clustering rather than on the specific mechanism of pion production (like higher-mass resonances [7,8]) In order to see, in a simplified manner, what the problem is, let us write the laboratory energy E L of nucleus A aswhere E N is the average nucleon energy. If one assumes that a collision results from the superposition of nucleon-nucleon collisions and ignores Fermi momentum and binding energy, the threshold kinetic energy per nucleon to produce, say, a pion is,This is the free nucleon threshold. In nuclear matter one may have nucleons with energy above E N and the threshold becomes lower than (2). The lowering of the threshold (2) can be very easily visualized if one allows for clustering of nucleons. If α ≤ A nucleons of nucleus A -with mass number A-collide with one nucleon of nucleus B -with mass number B -(or vice-versa) the threshold energy per nucleon becomes,This is, of course, the threshold for free nucleon-nucleus collisions. As experimental production of pions occurs even at energies bellow m π , this requires clustering from both nuclei, α ≤ A and β ≤ B,with the absolute thr...