LETTERS TO THE EDITOR 835 function, where € is the IT kinetic energy divided by 50 Mev. Since we only have knowledge of the shapes of /o and /i in the region of small e, they must be given the same normalization in this region if the test is to be a comparison of one shape against another:where p(e) is the phase space. For /o=l, fi-SA2e. The relative probability of spin 1 to spin zero is thenPo *-*Met) It should be emphasized that this factor of 10 40 against spin 1 is an underestimate of the odds against spin 1 (or any odd spin). This is because above 10 Mev the shape /i is normally expected to depart appreciably from the data (the data are close to isotropic in e). The most generous one can be toward spin 1 is to assume that j\ is a perfect fit to the data above 10 Mev and that it goes as p 2 from 0 to 10 Mev. This is exactly what was done in Eq. (1). A momentum dependence faster than n~2 would give an even worse fit. The above procedure was repeated using only the data below 5 Mev (31 events). This should be a weaker test because now f± is permitted to be a good fit all the way down to 5 Mev. In this case the normalizations are /o=l and /i=16.7e which gave odds of 24 to 1 against /i.The effect of Coulomb enhancement should not appreciably alter the p 2 dependence of spin 1. As has been pointed out, 11 -12 the entire low-energy region of the spectrum might be boosted up as much as 10%. In fact one might expect the enhancement factor to increase with p in this energy region. 13 Such a Coulombcorrected /i would give a worse fit than the /i used in Eq. (1). When the energy region 0 to 10 Mev is considered by itself, the relativistic corrections are negligible and have not been made. Over this region the nonrelativistic phase space is proportional to the relativistic phase space to within 1%. The result given by Eq. (1) still holds if one employs an arbitrary mixture of both parity states. Such mixing would make odd spin an even worse fit. Corrections for detection bias would also increase the odds against spin 1.The conclusion is that it is extremely unlikely that the r meson has spin 1 or any odd value. Spins 0 and 2 are both quite consistent with the data and there seems to be no way to distinguish them by means of a Dalitztype analysis alone. 2 If there is only one K meson, it most probably has either spin 0 or 2. The continued absence of the mode K + ->7r + +y is evidence against spin 2. 14 '* Research supported in part by the Office of Naval Research and the U. S. Atomic Energy Commission. 1 R. Dalitz, Phil. Mag. 44, 1068 (1953). 2 Orear, Harris, and Taylor, Phys. Rev. 102, 1676 (1956). 3 Feld, Odian, Ritson, and Wattenberg [Phys. Rev. 100, 1539 (1955)] give 55 r + mesons and in a later private communication give 25 additional r + mesons analyzed for low-energy pions only. 4 R. P. Haddock, Nuovo cimento 4, 240 (1956) gives 100 r + mesons. 5 Biswas, Ceccarelli-Fabrichesi, Ceccarelli, Cresti, Gottstein, Varshneya, and Waloschek, Nuovo cimento 3, 825 (1956) give 87 r+ 6 Brene, Hanson, Hooper, and Scharff, Nuovo ...
