Abstract
EDINBURGH
Royal Society, March 4. A. M. COCKBURN: The geology of St. Kilda. The islands of the St. Kilda group appear to represent the peripheral relics of a larger complex of intrusive igneous rocks perhaps some seven miles in diameter. The intrusions are sheet-like and, in order of age, consist of gabbros (including olivine-eucrite), coarse- and fine-grained dolerites, basalts and three granophyres. On St. Kilda these masses, the eucrite excepted, are inclined outwards from the supposed centre of intrusion, suggesting ring-dyke structure. A very abundant series of variously dated, thin, gently inclined sheets are suggestive of cone intrusion. Arcuate structure, however, has not been observed in the field owing to the fragmentary nature of the exposures. J. L. BHADUBI: The anatomy of the adhesive apparatus in the tadpoles of Rana afgkana, Gunther, with special reference to the adaptive modifications. A histological description of the cement organs is given and attention is directed to the ‘brush-border’ fringing the gland cells. The muscles of the disc are described and their homologies discussed. The diaphragm is shown to be modified; not only has it a thickened tendinous ridge in its central part, but in addition it possesses two distinct apertures for the passage of the diaphragmatobranchialis medialis muscles. An account of the histology of the skin of the disc is given. The cornification, and later tuberculation, of the posterior region of the disc is shown to be correlated with the habits and habitats of the tadpoles. The rim of the sucking disc encloses a lymph space which is considered to be another adaptive modification for counteracting the pressure of the rushing current. A. C. AITKEN: Least squares and linear combination of observations. The paper deals with the equivalence, in practical outcome, of two different approaches to least squares, one based on the assumption that errors are normally distributed in such a way as to give the observations the greatest probability, the other on the postulate that the values to be adopted should be weighted means of the observations, with smallest standard error. Dr. W. F. Sheppard had proved this equivalence for representation by polynomial curves. The present paper confirms it for general functional representation, as well as for the case of correlated errors.
Article PDF
Rights and permissions
About this article
Cite this article
Societies and Academies. Nature 135, 558–560 (1935). https://doi.org/10.1038/135558a0
Issue Date:
DOI: https://doi.org/10.1038/135558a0