As part of their business, Lensrentals carry quite a selection of interchangeable-lens camera bodies and lenses from various manufacturers, and they don't just stock a few copies of each item. It's a pretty well-established fact of life that slight (and sometimes, not so slight) variations occur across a production run, the camera industry being one where product pricing is a fairly important part of the equation, whether for enthusiast or professional products. As Lensrentals CEO Roger Cicala notes in a recent blog post, his company typically has dozens (and sometimes dozens upon dozens) of copies of any given lens or camera body available for rental, and that makes him uncommonly familiar with the issue of product variance.
For the average man--or product reviewer--on the street, it's just not feasible to test out more than a handful of samples of any given product before offering one's opinions, as the cost would quickly become prohibitive even for products aimed at enthusiasts, let alone professional gear. (Although with that said, early last year our sister site SLRgear.com published a couple of articles looking at sample variations with prime lenses from Canon and Nikon, with Lensrentals' assistance.)
For Lensrentals, though, it's a different matter--and that's thanks not only to the number of copies they have on hand, but also to the fact that they routinely test each product after every rental to ensure it performs to spec before going out to the next customer. All that testing not only keeps Roger and his technical staff busy, but it has led them to build up a wealth of information on product variance, be it across a production run, between individual batches (at least, where the manufacturer provides a way to ascertain this), or over the lifetime of a given item.
Thankfully for us, Roger's recognized the value of this mountain of data, and taken the time to turn it into a very interesting article looking at the issue for both lenses and camera bodies. The post includes a number of graphs that nicely demonstrate the difference between normal sample variance, and a product that's genuinely starting to demonstrate real-world performance issues. Entitled "Notes on Lens and Camera Variations", the article is well worth a read for anyone who's obsessed over getting the sharpest possible images from their SLR or system camera, as are a couple of related articles which Roger's published over the last couple of years: "'This Lens is Soft', and Other Myths", and the followup "'This Lens is Soft', and Other Facts". Interesting stuff!