for by varying the amplifier gain. Lead shielding between the two counters minimized scattering between the crystals. Figure 1 shows the single-crystal pulse-height distribution obtained in channel I for In 111 . The two photoelectron peaks are associated with the 172-and 247-kev gamma-rays. Figure 2 shows the zero-delay coincidence spectrum obtained when channel II was adjusted to transmit only the 247-kev peak with a large slit width and channel I ranged over the entire spectrum. A coincidence peak is seen at the position of the 172-kev gamma-ray since the time resolution of the coincidence circuit is insufficient to prevent true coincidences at zero delay. The ratio of true to chance coincidences at the peak is about thirty to one. The introduction of a 0.1-Aisec delay in channel I alone increased the coincidence rate by a factor of two, whereas the same delay introduced only in channel II practically eliminated true coincidences. This shows that the 247-kev gamma-ray is delayed with respect to the lower energy one and establishes directly the correspondence between the Cd m 247-kev level and the 247-kev gamma-ray associated with the decay of In 111 . I T is well known that the zenithal distribution of the integral meson spectrum at sea level obeys the empirical law Ie-Iocos x d } where 1$ is the intensity of cosmic-rays making an angle 0 with the zenith. The value of X is approximately two. It has been shown, 1 however, that this law does not apply for the differential spectrum, in particular, in the momentum range 300-410 Mev/c.In the present experiments a limited band in the spectrum was obtained by an arrangement of counters and absorbers such that the particles which stop in a block of lead are detected. In Fig. 1 the counters ABC define a beam which has passed through ababsorbers Pi, P 2 , and P 3 . P 2 and P 3 are each 5 cm thick. By ob-serving the particles which are registered in ABC but not in D t a band of the spectrum is selected which is fixed in position by P\-\-P 2 -\-Pz and in width by the thickness of Q\. At the same time a second band of longer range and of width fixed by Q 2 can be studied by observing particles which register in coincidence in A BCD but not in E. A group of counters G was included to exclude showers.The whole apparatus was mounted in a steel frame that could be rotated about a horizontal axis. For the results described here Pi+P 2 -\-Pz=lS cm of lead and Qi and Q 2 were each 7.5 cm thick.The results were corrected for the background and for the scattering in the absorber. These corrections will be described when a detailed account of the experiment is published. Table I shows the counting rate AIBD for particles stopped in the absorber Q\ and AI$E for particles stopped in Q 2 . Table II shows the values of X calculated from logr/log cos0, where r=Ald/Alo. Since the values of X are far from constant, the cos 2 0 law is not applicable. A much better fit is obtained if a function of the form r=(l -asin b e)is assumed. If log(X-r) is plotted against log sin0, the points fall on...

Experiments have been carried out at sea level to measure penetration of particles from local showers produced in lead as a function of the angular width of the cone of shower particles. The shower particles are assumed to be mesons and are observed by counter arrangements connected in coincidence and anticoincidence.The results show that the showers containing the more penetrating particles are those of wide angular spread. The multiplicity of penetrating particles per shower is estimated to be at least 3.7, and the curve of the angular distribution shows a maximum corresponding to a shower cone of a half angle of about 15°.

Sommaire.-Construction et conditions de fonctionnement d'une chambre à diffusion permettant l'observation continue des trajectoires des particules ionisantes. André VOISIN, décédé le 1er décembre 1952, disparait en pleine force de l'âge et laisse un vide non seulement parmi les siens, parmi ses amis et,camarades, mais aussi parmi les jeunes physiciens français. Le travail que nous publions aujourd'hui est son oeuvre. Nous n'avons fait que l'achever d'après les indications qu'il n'a cessé de nous donner durant les semaines, où la maladie qui devait l'emporter le contraignait de garder le lit. En rédigeant son dernier travail, je tiens à rendre hommage ici à la mémoire de l'excellent collaborateur et ami,' au camarade charmant et dévoué et au physicien de valeur que fut André VOISIN.

average (Ap/p) pur e=10.2±0.4XlO~5/atmos. was obtained for the pressure coefficient of resistance of germanium at 300°K. Assuming a compressibility 1.4XlO~6/atmos. and a coefficient of linear expansion 8XlO-fi /°C leads to a value AJE G /Ar=-0.87X10~4 ev/°C. This result is, with one exception, in excellent agreement with calculations from other experiments. Shockley and Bardeen 1 found for this constant the value -0.9X10~4 ev/°C using data on the mobilities of holes and electrons; their result is proportional to m c 6/4 . Johnson and Fan 2 have estimated from data on the temperature dependence of the Hall effect a value -1.1 X 10~4 ev/°C. Correction of this result by subtraction of the contribution of the lattice vibrations computed by Fan 3 leads to the value -1.0 X 10~4 ev/°C for the contribution of the volume effect alone. Only the latter value is comparable to the results of the pressure measurements; the uncorrected result of Johnson and Fan, which depends on (m e mh) zii , includes both the volume effect and the change in lattice vibrations which accompany a change in temperature. Fan, 4 using data from optical absorption measurements found the quite different value -4.0X10" 4 ev/°C.

Les mesures ont port6 sur la composante ptne'trante qui est surtout compos6e de mdsons p au niveau du sol. Beux bandes contigues ont 6th s6lectionn6es dans le spectre difi6rentiel; les impuisions des particules enregistrkes s'ktendent de 300 B $10 hlev./c et de 410 B 510 Mev./c pour des mdsons. La distribution de leur intensit6 a kt6 6tudike pour les angles au z6nith compris entre 0" et l$OO. Les rdsultats indiquent que la loi qui r6git la distribution &nithale des particules 6tudiCes est diff6rente de celle qui exprime la distribution z6nithale du spectre int6gral.(2) Absorption dans les parois des compteurs.--Cette absorption est calculable et represente environ 10yO du fond continu total. D'ailleurs, elk n'est pas un effet parasite et provoque de vkritables anti-coincidences.Can. J. Phys. Downloaded from www.nrcresearchpress.com by YORK UNIV on 11/20/14For personal use only.Etre manquees par les compteurs D. I1 est improbable qu'un kvenement de ce genre se produise avec une frequence suffisante pour alterer le resultat des mesures. RIais il n'en n'est pas de msme en ce qui concerne la diffusion dans le second &ran differentiel. Dans ce cas, meme une faible deviation des partiCan. J. Phys. Downloaded from www.nrcresearchpress.com by YORK UNIV on 11/20/14For personal use only.Les mesures par la rnkthode des anti-coincidences sont delicates dans le cas des angles 8 2 90" car, dans ces dernieres mesures, la tdlescope est aussi dCclench6 par les particules ayant des directions opposbes Q celles des particules Can. J. Phys. Downloaded from www.nrcresearchpress.com by YORK UNIV on 11/20/14For personal use only.