12.07.2015 Views

1 Studies in the History of Statistics and Probability ... - Sheynin, Oscar

1 Studies in the History of Statistics and Probability ... - Sheynin, Oscar

1 Studies in the History of Statistics and Probability ... - Sheynin, Oscar

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

values <strong>of</strong> (2.5) are essentially less than those <strong>of</strong> <strong>the</strong> Kolmogorovstatistic (2.4). Therefore, when apply<strong>in</strong>g <strong>the</strong> Kolmogorov distributionfor (2.5), we will widen <strong>the</strong> boundaries <strong>of</strong> <strong>the</strong> confidence region <strong>and</strong>thus admit <strong>the</strong> hypo<strong>the</strong>sis <strong>of</strong> normality more <strong>of</strong>ten than proper.And so, <strong>the</strong> careful practical application <strong>of</strong> <strong>the</strong> Kolmogorov test <strong>in</strong><strong>the</strong> most elementary (<strong>and</strong> <strong>the</strong>refore most common) situation isimpossible. That criterion helps <strong>in</strong> those cases when many suchexamples are available which were already tested by some statisticalcriteria <strong>and</strong> we wish to secure a general po<strong>in</strong>t <strong>of</strong> view concern<strong>in</strong>g <strong>the</strong>irnumerous applications.Let us pass now to <strong>the</strong> essence <strong>of</strong> <strong>the</strong> problem on <strong>the</strong> confirmation<strong>of</strong> <strong>the</strong> Mendelian laws. Here is <strong>the</strong> classical situation. Some <strong>in</strong>dicationhas two alleles, A (dom<strong>in</strong>ant) <strong>and</strong> a (recessive). Two pure l<strong>in</strong>es withgenotypes AA <strong>and</strong> aa are taken <strong>and</strong> compulsorily crossed. A hybridwith genotype Aa emerges with its phenotype correspond<strong>in</strong>g to<strong>in</strong>dication A. Then a second generation is obta<strong>in</strong>ed under free cross<strong>in</strong>g.When admitt<strong>in</strong>g <strong>the</strong> hypo<strong>the</strong>sis <strong>of</strong> absolute r<strong>and</strong>omness <strong>of</strong> <strong>the</strong>comb<strong>in</strong>ations <strong>of</strong> <strong>the</strong> gametes, <strong>the</strong> probability <strong>of</strong> <strong>the</strong> occurrence <strong>of</strong>genotype aa is 1/4. Only <strong>in</strong>dividuals with genotype aa reveal<strong>in</strong>dication a <strong>in</strong> <strong>the</strong>ir phenotype so that <strong>the</strong> probability <strong>of</strong> its occurrenceis also 1/4. And so, if <strong>the</strong>re will be n <strong>in</strong>dividuals <strong>in</strong> <strong>the</strong> secondgeneration, <strong>the</strong> number <strong>of</strong> occurrences <strong>of</strong> <strong>in</strong>dication a <strong>in</strong> <strong>the</strong> phenotypemay be considered as <strong>the</strong> number <strong>of</strong> successes µ <strong>in</strong> n Bernoulli trialswith probability <strong>of</strong> success p = 1/4.This is <strong>the</strong> simplest case <strong>of</strong> <strong>the</strong> Mendelian law. Vast experimentalmaterial had been collected up to 1940 from which it was seen that <strong>in</strong>many cases such a simplest law was <strong>in</strong>deed obeyed. Essentialdeviations (perhaps connected with a differ<strong>in</strong>g survivorship <strong>of</strong><strong>in</strong>dividuals <strong>of</strong> different genotypes <strong>and</strong> o<strong>the</strong>r causes) was also revealed.The school <strong>of</strong> Lyssenko had been attempt<strong>in</strong>g to prove that that lawwas not work<strong>in</strong>g. To atta<strong>in</strong> that aim, experiments were carried out, <strong>in</strong>particular by Ermolaeva (1939). They were peculiar <strong>in</strong> that <strong>the</strong>material was considered not from all <strong>the</strong> <strong>in</strong>dividuals <strong>of</strong> <strong>the</strong> secondgeneration taken toge<strong>the</strong>r, but separately for families. It is better toexpla<strong>in</strong> <strong>the</strong> mean<strong>in</strong>g <strong>of</strong> that term by an example. In experiments withtomatoes a family is consist<strong>in</strong>g <strong>of</strong> all <strong>the</strong> plants <strong>of</strong> <strong>the</strong> secondgeneration grown <strong>in</strong> <strong>the</strong> same box. Each box is sown with seeds takenfrom <strong>the</strong> fruit <strong>of</strong> exactly one plant <strong>of</strong> <strong>the</strong> first generation. Theseparation <strong>in</strong>to families occurs quite naturally.However, Kolmogorov (see above) showed that Ermolaeva’s mostnumerous series <strong>of</strong> experiments can be expla<strong>in</strong>ed exactly by <strong>the</strong> mostelementary Mendel model. Suppose that for k families number<strong>in</strong>g n 1 ,n 1 , ..., n k <strong>the</strong> number <strong>of</strong> manifested recessive alleles was µ 1 , µ 2 , ..., µ k ,<strong>the</strong>n <strong>the</strong> classical De Moivre – Laplace <strong>the</strong>orem [prov<strong>in</strong>g that <strong>the</strong>b<strong>in</strong>omial law tended to normality] leads to <strong>the</strong> normed magnitudes* µi− nip 1 3µi= , p = , q = 1− p =n pq 4 4ihav<strong>in</strong>g approximately <strong>the</strong> st<strong>and</strong>ard normal distribution N(0, 1); <strong>the</strong>precision <strong>of</strong> approximation is quite sufficient for n i <strong>of</strong> <strong>the</strong> order <strong>of</strong>102

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!