On the source of parochialism in Case Transmission I. Although case agreement on adjectival secondary predicates has inspired linguists of different theoretical persuasions for at least the past 25 year, the cross-linguistic aspect of it still remains a mystery. Icelandic, Czech and Polish are three languages which display striking differences concerning case agreement in infinitival complements. To the extent that case agreement with the Subject in raising verbs is expected on theoretical grounds, case agreement in control environments has constituted a problem for the Case-based account of the distribution of PRO for two reasons: (i) if floating quantifiers can agree with PRO for quirky case, then PRO must be casemarked (cf. e.g. SigurÏsson, 1991), (ii) if predicative adjectives agree with PRO's antecedent then PRO must somehow mediate in case-transmission (cf. e.g. Franks, 1995), as agreement is clausebounded and the antecedent has never been in the embedded clause. On the other hand, movement approaches to (Obligatory) Control (cf. Hornstein, 1999 and subsequent work) circumvent the problem in (ii) while facing a different challenge related to (i), i.e. how is movement possible out of a case-marked position. If, however, C/case does not drive movement and is instead epiphenomenal spell out of various sizes of a universal nominal functional sequence (fseq ) resulting from the argument moving through subsequent positions in the verbal fseq (i.e. Starke's peeling), the movement approach becomes unproblematic. II. The control paradigm I focus on is as follows: in both Subject Control and Object Control Icelandic displays the option of the predicative adjective agreeing with NOM, ACC and DAT Subjects and NOM, ACC and DAT Objects (I illustrate the crucial ACC and DAT Object agreement only in (2a) and (2b)), apart from the more generally available non-agreeing Nominative adjective (cf. SigurÏsson, 2002). In Czech, on the other hand, agreement with Dative Objects (4b) is excluded (note that Dative Subjects are arguably absent), whereas agreement with Nominative and Accusative ECM Subjects ((3a) and (3b)) and Accusative objects ((4a)) is allowed, alongside the Nominative option (cf. PrzepiÑrkowski and Rosen, 2004).1 Finally, the agreeing option is most restricted in Polish, where agreement with Accusative or Dative objects ((5b) and (5c)) is excluded, inducing instead the generally available Instrumental case on the adjective (cf. Bondaruk 2008). In Polish it is only the Nominative subject that can control case agreement on the adjective ((5a)). III. Following the idea in Taraldsen (2006), I will argue that there is a crucial boundary in the nominal fseq , call it Xn P, which is opaque for the percolation of features required for the relevant kind of agreement. Cases which spell out nominal structures more deficient than Xn P will result in predicative agreement, whereas for more `unpeeled' arguments agreement would be blocked. The particular lexical specification of case markers for all the three languages is in (1). (1) a. b. c. Icelandic: [Xn P [DAT [ACC [NOM ]]]] Czech: [DAT [Xn P [ACC [NOM ]]]] Polish: [DAT [ACC [Xn P [NOM ]]]]