In case it's not well-known, much of the contracted out PRO / TNA mass census transcription 1841-1891 (ex 1881) and 1891, was done on the Indian sub-continent, checked by youngsters who were inadequately supervised and the results lamentlably not overseen as a glance at some of the nonsense shows.
Youth and inexperience should not be let loose on such work.
It is clear that some transcribers were more than competent but that the PRO was party to this scramble is a scandal which it would take a lot of money to sort out.
(Was any attempt made to read faint entries under a UV lamp?).
Having spent 5 years organising a team to transcribe 1851 for my county FHS, transcribing thousands checking all half million - while making no guarantee of accuracy, I reckon to have a fair knowledge of what to look for. I offered to check the PRO 1851 but was sent packing
Being a bit slow to cancel my sub to ancestry.co.uk, I spend a lot of time seeking out errors and like dundurn have made many corrections and suggestions - ca 2000 so far and am fed up with the messages of thanks!
Presumably this sort of thing is being done on other providers sites so it is a complex muddle.
Currently working on Do and D which signified Ditto.
Any surname which looks nonsense, more often than not will be found to be just that.
A quick look at the surname FOSE in FreeBMD shows comparatively few - but look at ancestry's line-up and the images, where the final 'x' looks more often than not like 'se'.
That's quite enough from me, except to say if you can't find what you wnat it may well be lurking somewhere, but don't overlook missing booklets, (eg 1871 Stepney, is it? and one in Battersea).