I wanted to update the wiki article on watermills and I started experimenting with them.
My current theory is that somehow the watermill is sending a separate energy packet for each block of water around it. This means that even a short length of tin cable is deadening the packets. However, when I connect a batbox up with a single tin cable I get what appears to be full packets. When I tried measuring it with the EU meter I got odd results, pics below.
So I'm thinking is that the watermill is checking the enet for a destination for each pulse* of the watermill and not finding them due to the loss on such a small packet and not sending. If it finds a valid destination it groups the pulses into one packet and then sends.
*energy generated per block of water - not sure what to call it
Nevermind, I made a very noobish mistake. I connected the watermill up to the output of the batbox.