How long should an astronomical paper be to increase its Impact?

Abstract: Naively, one would expect longer papers to have larger impact (i.e., to be cited more). I tested this expectation by selecting all (~30,000) refereed papers from A&A, AJ, ApJ and MNRAS published between 2000 and 2004. These particular years were chosen so papers analyzed would not be too "fresh", but at the same time length of each article could be obtained via ADS. I find that indeed longer papers published in these four major astronomy journals are on average cited more, with a median number of citations increasing from 6 for articles 2-3 pages long to about 50 for articles ~50 pages long. I do however observe a significant "Letters effect", i.e. ApJ and A&A articles 4 pages long are cited more than articles 5-10 pages long. Also, the very few longest (>80 pages) papers are actually cited less than somewhat shorter papers. For individual journals, median citations per paper increase from 11 for ~9,300 A&A papers to 14 for ~5,300 MNRAS papers, 16 for ~2,550 AJ papers, and 20 for ~12,850 ApJ papers (including ApJ Letters and Supplement). I conclude with some semi-humorous career advice, directed especially at first-year graduate students.