I was rather shocked when hearing in a conversation that journals sometimes have explicit limits on the number of references that can be included in one article. I understand that there are space concerns for print publications, and that word/page limits, as well as limits on the numbers of figures and tables are common. But placing limits on the number of references strike me as rather close to impinging on the integrity of the articles themselves (would anyone support a motion to limit the number of authors?).
Question: What are some reasons that have led journals to limit the number of references?
Clarification: I am not asking whether this is a good idea in general. (I think it is silly, but that's just my opinion.) I am not asking whether this can have some potentially good effect on the quality of scientific writing. I am specifically asking whether there was an official explanation ever issued on the part of the publishers explaining this rule, or whether there was some event (say, an abuse in the form of many gratuitous references or an observed trend for the average number of references to keep growing if not otherwise checked) or some strong personality (famous editor-in-chief X) that led to these kinds of policies being formulated.
For example, Applied Physics Letters has limits on number of words (with some conversion factor applied also to figures and tables), but (in my opinion, rightly) excludes the title, the author list, and the list of references from the limit.
For examples of journals (from various publishers) that have limits on number of references:
- Nature "strictly enforces" a limit of 50 for articles and 30 for letters. (Science, I note, does not for research articles. For review articles the limit is 100.)
- Earth and Planetary Science Letters limits to 50.
- Blood "recommends" a limit of 100 references, though I don't know if this is a hard limit or not.
- Journal of Clinical Ontology limits to 10 for "correspondences" and a "suggested limit" of 150 for "research articles".
- Journal of Genetics (more reasonably, in my opinion), have different limits for different types of articles. But notably for research and review articles there are no limit to references: the reference number limit applies to shorter submissions such as correspondence, commentary, or "research notes".
Some additional information:
(TL;DR: Nature put in their current policy sometime between the years 1986 and 1988, with no reference limits prior. Did something happen in the 80s?)
I went down to the library and looked at the old issues of Nature (not because I want to single them out, but because the library happen to have all the issues since the 20s available on the stacks). After some binary search I found that in December 1986 the instructions to the authors don't mention a limit to references:
GUIDE TO AUTHORS
Authors should be aware of the diversity of Nature's readership and should strive to be as widely understood as possible.
Review articles should be accessible to the whole readership. Most are commissioned, but unsolicited reviews are welcome (in which case prior consultation with the office is desirable).
Scientific articles are research reports whose conclusions are of general interest or which represent substantial advances of understanding. The text should not exceed 3,000 words and six displayed items (figures plus tables). The article should include an italic heading of about 50 words.
Letters to Nature are ordinarily 1,000 words long with no more than four displayed items. The first paragraph (not exceeding 150 words) should say what the letter is about, why the study it reports was undertaken and what the conclusions are.
Matters arising are brief comments (up to 500 words) on articles and letters recently published in Nature. The originator of a Matters Arising contribution should initially send his manuscript to the author of the original paper and both parties should, wherever possible, agree on what is to be submitted.
In October 1988 the guidelines it became almost the same as present day.
GUIDE TO AUTHORS
Please follow these guidelines so that your manuscript may be handled expeditiously.
Nature is an international journal covering all the sciences. [...] Because of the competition for space, many of the papers submitted for publication cannot be accepted. For this reason, and because brevity is a great assistance to readers, papers should be as brief as is consistent with intelligibility. Please note that one printed page of Nature, without diagrams or other interruptions of the text, has fewer than 1,300 words.CATEGORIES OF PAPER
Review Articles survey recent developments in a field. Most are commissioned, but suggestions are welcome in the form of a one-page synopsis addressed to the Reviews Coordinator. Length is negotiable in advance but must not exceed six pages of Nature.
Articles are research reports whose conclusions are of general interest and which are sufficiently rounded to be a substantial advance in understanding. They should not have more than 3,000 words of text or more than six display items (figures and tables) and should not exceed more than five pages of Nature. [...] There should be fewer than 50 references.
Letters to Nature are short reports of outstanding novel findings whose implications are general and important enough to be of interest to those outside the field. Letters should not have more than 1,000 words of text or more than four display items and should not occupy more than two pages of Nature. The first paragraph should describe, in not more than 150 words, the origins and chief conclusions of the study. Letters should not have subheadings or more than 30 references.
(I don't have the exact date of the switch, since only some of the issues in the library came from the original magazines; others came from bound reprints ordered from the publisher which stripped out pages like these. So in particular I found no "Guides to Authors" in the 1987 issues I had access to.)
Back in the 20s and 30s most of the items in Nature had no references whatsoever. By the 50s and 60s we start seeing articles more in the form of what we expect today, but the number of references are generally not too many. Even in the 70s and the 80s (before the change of rules) the majority of the articles do abide by the modern rules, with occasional exceptions.