I'm finding this one tricky too. It's similar to a case I was working on for a friend - I never found the 'first' marriage and admit defeat on that one. Could be he never actually married the lady. Perhaps she was already married. Perhaps it all happened abroad. Perhaps he wasn't widowed and it was a bigamous marriage. Interesting snippet from this previous case was in the 1939 Register he described himself as widowed when he wasn't, and I wonder if this was a device to avoid awkward questions.
And his birth - looks like you have the certificate because you specify the exact date - may I just check the rest of the info - place of birth, parents names, occupations, informant etc? In working around this one, looks like the place of Lingfield, Surrey was in the registration district of East Grinstead, Sussex, which is all set to confuse.
I have found a candidate for the first marriage, but the father's name was James and not William. I was going round in circles big time last night! I kept coming across various members of the Bonwick family - is that a name you know?
May I be extremely picky for the moment and ask where your information comes from?
I have found the online transcription of James' marriage to Lucy Todman - do you have the certificate? I have known the father's name transcribed wrongly, although it's clear enough on the image. Even so, there are no rules as to what goes into the father's name and occupation boxes and it could be a father figure, and not the biological father.
I'll have another go this evening and create a tree on Ancestry for the purpose. Sometimes the hints throw up something I miss, and it's easier to see what other researchers have done, be it right or wrong...