This appears to be an error in citation propagated by copying of the (incorrect) citation unread. The work is real, On Marvellous Things Heard, "Mirab." being an abbreviation for the Latin title De mirabilibus auscultationibus. The work is traditionally attributed to Aristotle, but is no longer believed to have been written by him. However, the particular reference is wrong. The reference should be to Aristotle, Mirab., 110 (not 117):

Among the Peucetini they say that there is a temple of Artemis, in which is dedicated what is called the bronze necklet, bearing the legend "Diomedes to Artemis." The story goes that he hung it about the neck of a deer, and that it grew there, and in this way being found later by Agathocles, king of the Siceliots, they say that it was dedicated at the temple of Zeus.

according to the version at http://penelope.uchicago.edu/Thayer/E/Roman/Texts/Aristotle/de_Mirabilibus*.html

Narrative 117 is unrelated, and must be wrong. Also note that this passage is cited with respect to a temple of Minerva in Apulia, but the passage mentions a temple of Artemis in Apulia.

The copying of references without reading is nothing new. M.V. Simkin and V.P. Roychowdhury, "Read before you cite!", Complex Syst. 14 (2003) 269-274 (available here) suggest that about 80% of references are copied without reading. Given this, studies finding error rates in references of 26% and 38% (even 45.8%!) are not so surprising (and it's the propagation of errors in reference lists that allows the estimation of the frequency of copying without reading). The problem of errors in references is many decades old (e.g., Evans, JT; Nadjari, HI; Burchell, SA, "Quotational and reference accuracy in surgical journals. A continuing peer review problem.", JAMA. 1990; 263: 1353-1354. There has even been work on the history of the problem, with attempts to measure changes over time in error rates.