Right now, I'm writing a review for myself of very related theories in two different materials systems. However, until very recently, there was fairly little overlap between the scientists who studied these two systems, and it shows in reviewing these papers. In one system, they started with empirical fits of data, and drifted to a more precise mathematical approach, whereas the other material started with math and has slowly been correcting to account for experimental results. The most interesting part is when both sides come up with the same result, and just have a different name.
It reminds me of the paper several years ago when medical students rediscovered integration.
Materials science is one of those boundary fields, with significant overlap with a number of fields, so I've always been encouraged to look at papers, even from unfamiliar journals. I've also made an effort to take classes outside of my departmental comfort zone (which has been rather helpful during this review). It also tends to mean I do a lot more reading than my friends in say, EE.
How many times have we reinvented wheels because we only read what we thought was directly relevant to our subfield?