What statistical book is your preferred one to give light to the main problems of study in Statistics, and why? They could be of all branches of knowledge but with good contents in this scientific speciallity of Statistics.
It depends what your needs are. The preferred book for graduates in statistics programs will likely be different than undergraduates in social sciences, and these differ from what a philosopher of science might want. Please provide more details.
'Statistics' by William Hays. He explains everything starting from the use and mean of the sum sign and set theory from the very beginning up to analysis of covariance in a easy way. Hays uses many formulas including proofs and providing therefore a deep understanding of statistics. It has 1100 pages.
If we're talking all time, some people might suggest Fisher's 1922 paper (http://www.unc.edu/depts/case/BMElab/MPNcalculator/Fisher_pt1921.pdf), though there will be lots of people these days who disagree with some of what he says. Again, thinking all time (but in that they talk about statistics over a long time) books that I see on many statisticians from different areas bookshelves are Stigler's (http://www.hup.harvard.edu/catalog.php?isbn=9780674403413) and Hacking's (http://www.cambridge.org/us/academic/subjects/philosophy/philosophy-science/emergence-probability-philosophical-study-early-ideas-about-probability-induction-and-statistical-inference-2nd-edition) history books.
Cochran, William G. (1977). Sampling Techniques, 3rd edn. Wiley. New York.
It was (in the three editions) the first complete mathematical book on sampling of finite populations, those that exist in reality to be studied with guaranties in much cases. Other addressed book to the objective statistical inference and on sampling of finite populations is:
Ruiz Espejo, Mariano (2013). Exactitud de la Inferencia en Poblaciones Finitas. Bubok. Madrid.