In an attempt to study details of cosmic-ray nuclear interactions a large cloud chamber containing eight \textonehalf{}-inch lead plates was operated in conjunction with a 134-counter hodoscope placed directly below the chamber, at Echo Lake, Colorado (altitude, 3260 m) and at Syracuse, New York (altitude, 220 m). Showers produced in the chamber were detected by counter trays above and below the chamber, producing a masterpulse which triggered both the chamber and the hodoscope. It was then possible to follow in the hodoscope the further development of the nuclear cascade started in the chamber, and to study for each event both the individual features of the interaction from the cloud-chamber photograph, and, at least statistically, the properties of the secondaries from the hodoscope record. In particular, the "integral path length" of the nuclear cascade in the approximately 600 g/${\mathrm{cm}}^{2}$ lead absorber of the hodoscope, is a measure of the energy of the primary particle. It is concluded that most of the primaries recorded had energies between 1 and 10 Bev. The analysis of about 400 shower pictures shows no appreciable difference between the events observed at the two locations. Both show a variation of the primary mean free path from about 190 g/${\mathrm{cm}}^{2}$ to 160 g/${\mathrm{cm}}^{2}$ over the energy interval covered. The statistics are not good enough to decide whether the mean free path of the secondaries also varies with the primary energy; a mean value of 250 g/${\mathrm{cm}}^{2}$ was observed. The ratio of the number of charged mesons to the number of protons increases from about 0.4 to 2.1, while the ratio of neutral to charged mesons is found to be 0.55\ifmmode\pm\else\textpm\fi{}0.1 for low multiplicities, and its lower limit for higher multiplicities is 0.36\ifmmode\pm\else\textpm\fi{}0.1. A distribution function was derived for the total shower multiplicities as a function of the primary energy. For low primary energies, small multiplicities are favored; but with increasing primary energy the distribution function becomes rather flat. The primary energy spectrum found agrees satisfactorily with the observations of other authors.