Not quite true, Sandra, despite the fact that this myth is bandied about all over Ancestry!
Registration of births was not in any way optional after 1 July 1837. The intention of the 1836 Act for registering Births, Marriages and Deaths in England was to enable a "complete" register of these events to be maintained. It was a pretty crummy piece of legislation (probably because it was brought in in the face of considerable opposition) and the mechanics of the registration procedure were not particularly well thought out. The main difference between how it worked then and the new system brought in in 1875 was the responsibility for registration.
The 1836 act (which had to be amended three times before it even came into force!) appointed Registrars whose responsibility it was to seek out and register *every* baby born in their district. The parents' only responsibility was to provide the details when asked. Registrars were paid a fee for each birth registered, so they had a strong incentive to make sure they didn't miss any. However, in overcrowded urban parishes or widely spread out rural ones even the keenest, nosiest registrar would have been bound to miss a few.
The 1875 Act changed the system around. Registrars were now provided with offices and it became the duty of parents (or occupier of any dwelling in which a baby was born) to turn up at the office and register the child. It was still part of the Registrar's duty to "inform himself" of all births in his area and to require parents to register their babies if they didn't volunteer to do so.
The change in the system did not increase compliance with the law. In fact, registrations dropped slightly afterwards.
Failure to find a birth registration in the GRO index can result from a number of factors:
1. The birth was not registered. This is more likely in very crowded or very empty areas. It is unlikely to be the case if the parents were living in a settled community.
2. The birth was registered but does not appear in the GRO index. This almost certainly accounts for a small percentage of entries that were "lost in transit" between the original registration and the compilation of the currently available index. Local registrars had to send quarterly returns to superintendent registrars. They, in turn, sent returns to the GRO. The GRO then compiled an index (handwritten until the 20th century). Many of the handwritten indices were later replaced with typed versions. This multi-stage process gave plenty of scope for entries to disappear or be incorrectly processed.
3. The birth was registered but under a different name or in a different place from that expected. A mother may have travelled some distance to a relative for her confinement or, if she was very fortunate, went to a lying-in hospital some distance from home. Children of doubtful parentage may well have been registered under a false name.
Hope some of this helps