The paper discusses Tarski’s approach to quotation. It starts from showing that it is vulnerable to semantic inconsistencies connected with what is known as Reach’s puzzle, formulated in 1938 by a Czech logician Karel Reach. This fact gives rise to serious problems concerning the relation between the metalanguage and an object language. Moreover, the paper touches upon a historic aspect, pointing out that the problem at hand is discussed in the only paper signed up as Al. Tajtelbaum, i.e. Alfred Tarski’s original name. It argues that the puzzle reveals the importance of reopening the discussion on the understanding and limitations of deriving the metalanguage from an object language.