3/8". Tom et al have a deal of just using a dog snap for tender i believe. This isn't part of the connection; just a tender, so doesn't have to be up to the tensile standard of the support chain. i just kinda like my minimalist gear.
The 3/8 will be lower tensile than Sohner's; but it is in a dual legged position on a dual legged support to my massive 135# (dripping wet). But, also, in such tight bight's and round cords, the lower diameter will give less leveraged distance on the bent axis; so less tensile loss. Of curse; a flat (Tenex; the large white looks flat before splice?) would have even less loss; by being A) more flexible, so therefore less loss because the leverageing weakness against tensile / tensile loss comes from the resistance to bending X B) less 'height'/length on the bent axis of the bight; so less leveraged distance. (Read that as tensile strength loss = resistance to bend X Height(leveraged distance) at the bend)
Witch brings me to my main point(lessnessism); that one short coming of a splice; is the stiffness X 'height'(?) at bend that it would lend; so therefore any splice should be on straight/ non-bent portion of line (for purists)(?). i just like pushing the theory; to grasp it more; ain't trying to plough into anyone(especially when everyone is bigger'en me!).
But, anywho; i think a smaller diameter would be more 'discrete'/out of the way, serve as well, and could avoid the last shortcoming by allowing termination at both ends/ rather than splice...(don't mind me, i'z just a fool fer'dis chit!)