PROBLEMS IN COMPARATIVE PHILOLOGY 65 



of Philology, xxn, p. 261) and which were supported by Ebbing- 

 haus's observations (Zeitschrift f. Psych, und Physiol. d. Sinnesorgane, 

 xix, 1902, p. 142) showed * that cases where one isolated numeral was 

 associated with another were excessively rare. I am, therefore, 

 inclined to believe that the frequent juxtaposition' of these two words, 

 without reference to the likeness of their grammatical category, 

 lies at the bottom of the change. And, in general, it would seem as 

 if too little weight is given in our whole treatment of analogy- 

 formation to the associations due to frequent juxtaposition and 

 habitual combinations. We are too apt to take words singly and 

 treat them apart from their ordinary setting. If a word for " day " 

 grammatically influences the word for " night " (as Nachts after Tags 2 ) 

 is it not because they frequently occur in close proximity in the 

 sentence rather than because the two are associated by sense? 3 The 

 coherence of the members of a phrase in the spoken language is 

 much greater than is usually supposed, as is proved by those instances 

 (rare enough in the revised written texts) in which one member 

 of a common word-group may be observed to carry in its train its 

 mate, though the latter be not needed or be even disturbing. Interest- 

 ing cases of such "agglutinative association" (cf. the author's Lec- 

 tures on the Study of Language, p. 183, sec. 16) are given by Kemmer 

 (Die Polare Ausdrucksweise in der griech. Litteratur, Schanz' Beitrdge, 

 vol. xv, p. 2; 45, 50, 57). They are paralleled in English by such 

 phrases as Colonel Henry Watterson's: * "It (Life) is racy of the 

 soil, even as Punch in London is racy of English soil, a reflection of 

 the moods and tens es of the time, of the thoughts and fancies 

 of the people "; and in these passages from a letter: "As J. was out 

 till morning doth appear, mother and I talked till late "; " This 

 is my regular in the springtime gentle Annie feeling." Such 

 cases 5 are the morphological counterparts to those phonetic altera- 

 tions where a word either loses or gains an initial by too intimate union 

 with another word, as Meiseribuhl (from im Eisenbuhl, cf. Zt. f. d. 

 deut. Unterricht, xvn, 1903, p. 728), and which, for English, are very 

 exhaustively treated by C. P. G. Scott (Transactions of American 

 Philological Association, xxm, 1892, p. 179, xxiv, 1893, p. 89). 



1 Cf. also Ebbinghaus, Grundzilge der Psychologic, 2d ed., i, 1905, p. 704. The 

 objections of Marbe and Watts are discussed in American Journal of Philology, 

 xxvi, 1905, p. 95, note 1. 



2 In the same way Lathi noclu after diu, Bartholomae, IF, x, p. 13. 



* In this way Gothic haimos owes its feminine gender to its frequent connec- 

 tion with baurgs (Dieter, AUgerm. Dial. p. 571, sec. 330, 2, note 3), late Icelandic 

 f0tr its feminine gender to frequent connection with hendr (Dieter, ibid.) etc. 



4 In an editorial in the Louisville Courier-Journal, reprinted in Life, vol. xin 

 (1903), no. 1099, p. 479. 



8 Cf . also ' high and dry protectionist ' in the editorial of the New York Evening 

 Post for May 22, 1905, and similar instances collected by H. Willert, Archiv /. d. 

 Stud. d. neueren Spr. u. Lit., vol. cxi (= N. F. xi) p. 420. 



