Scientists love a good joke as much as anyone. They just happen to have very well-developed and incredibly nerdy splinter interests. The Journal of Irreproducible Results (JIR) was (and still is) a quixotic attempt to make a general satirical publication for scientists as a group; being for
serious formally-educated scientists, it was naturally published in the form of a journal.
JIR was founded in Israel in 1955 by virologist Alexander Kohn and physicist Harry J. Lipkin, who wanted a humor magazine about science, for scientists. While JIR has since passed through the hands of a number of editors and publishers, it has consistently published a mix of jokes, puns, cartoons, satirical essays, and general nonsense, all sciencey in nature. It will occasionally include bits of real, but amusing, research, but this is the exception rather than the rule.
JIR is perhaps the most famous/popular of the faux scientific journals, although ex-editor (1991-1994) Marc Abrahams went on to create and edit the popular rival publication, Annals of Improbable Research (he also created the Ig Nobel Prize). There are some samples of JIR publications available on-line, mostly available at their web-site, JIR.com, although many of the links to archived articles are currently broken. You may have seen the winner of a recent JIR reader contest floating around the internet; the One Graph to Prove all Scientific Theories. This graph is fairly representative of the sort of humor found in the journal. If this looks like your sort of fun, you can either subscribe to the journal, or buy one of their many anthologies.
JIR is currently edited by astronomer Norman Sperling, who has sworn to rejuvenate the publication. He has been working on this since 2004, and has had limited success. However, the JIR is something of a legend in geek circles (both computer and science), and is fondly remembered and/or sought after by many odd fans.