The measurement of linkage in heredity / by K. Mather.

  • Mather, Kenneth, 1911-1990.
Date:
[1938]
Catalogue details

Licence: Attribution-NonCommercial 4.0 International (CC BY-NC 4.0)

Credit: The measurement of linkage in heredity / by K. Mather. Source: Wellcome Collection.

    21/152 (page 7)
    exceeded or equalled once in twenty trials was found. The hypothesis is then rendered unlikely as it could account for such a family only once out of twenty times. When twenty families are grown it can account for one such family in each trial and is not unlikely. The level of probability chosen as indicat ing significant departure from hypothesis is simply the level at which the worker is willing to be misled. If, as is usual, the one in twenty level is taken, he will find that his supposedly real departures are. actually chance ones, once in twenty cases. If the one in a hundred level is taken he will be wrong, in calling the departure real, less often, but if a hundred such cases are taken he must expect to be wrong once. If this is constantly borne in mind the experi menter will set his levels of significance to suit his circumstances and will not be disconcerted when an apparently promising line of work comes to nothing because it was based on a false conclusion as a result of his test of significance misleading him. The only exception to this rule, in genetics, is when segregation is observed to occur in what should be, by the hypothesis, a homozygote. Even this cannot be considered as a complete exception, as the ' segregation ' could be the result of mutation, or error in the handling of the material. When the first and simplest hypothesis has been shown to be unlikely, another may be set up and tested. The new hypothesis may involve a para meter, a numerical quantity characterising the population, which must be estimated, e.g. the sup position that two genes are linked demands that an estimate of the recombination value be obtained before the hypothesis of linkage is sufficiently precise to be tested by observation. This involves the use of a method of estimation. When the hypothesis has been formulated precisely it may, in its turn, be tested against the observational data by a new test of significance. 2