The primary theoretical goal of the present study is to verify a hypothesis that fixedness of word combinations is notnecessarily connected with non-compositionality. Many constructions formed in accordance with the rules governing the cooccurrenceof their elements can nevertheless be retained in memory as separate units. Using large text corpora for theempirical data we are going to analyze the type of construction в том-то и весь N (that’s /just the whole N) which is realizedin a wide variety of tokens and demonstrate that some tokens of the construction can be so frequent that they can beconsidered to be cognitively entrenched units and are preserved in memory as separate units of the language. Such unitsshould be described as separate items of the lexicon. The practical task of the investigation is to refine our notions about thestructural peculiarities of в том-то и весь N (that’s /just/ the whole N) and its variant в том-то весь и N (that’s /just/ thewhole N).a We are also going to identify some regularities of distribution of fillers of the slot N in both variants of theconstruction and formulate corresponding rules for such distribution.