Hi xantos,
We first need to be sure that the painting has been 'published' at some stage during its life. If it hasn't, then copyright in it will still exist and you would be prevented from making copies of it without first gaining permission from the heirs of the artist, or more likely, the heirs of the person who commissioned the painting in the first place.
You say that the engraving which was made from the painting is not an exact facsimile of the painting. The degree to which the former differs from the latter would determine whether the printed version of the engraving constituted a copy of the painting which was made available to the public, ie 'published'. If the differences are minor then there probably won't be a dispute about this, since the law deems that if a substantial part of the original has been published with the permission of the artist, then publication of the copies will have activated the copyright, which then would subsequently have ceased at some later point, probably fifty years from the date of publication*. However there is a 'but' here. That involves the fact that the original was in colour whereas the engraving would have been in black and white. Based on some very old caselaw, this could mean that a substantial element of the original was not reproduced, namely the colour, meaning that publication might not be deemed to have occurred. It is therefore important to know whether the prints from the engraving were hand-cloured before being issued to the public. If they were then the argument about substantiality would probably fall away and we can assume that the painting has been published, even though the hand colouring may well differ from the original watercolour. If the prints were in black and white only, a question mark might remain.
On the assumption that no copyright remains in the original painting you can certainly go ahead and make new reproductions of it and sell the prints. The question of whether you will thereby create a new copyright is a fraught one on which academic opinion is divided. As the law currently stands, a work is only eligible for copyright if it is 'original' (see
section 1(1)(a) of the Copyright Designs and Patents Act 1988). This means that the work must be the intellectual creation of the author (see paragraph 37 of judgment of the Court of Justice of the European Union in the case of known as
Infopaq). Clearly mere mechanical copying, say by scanning or digital photography, would not meet this criterion. Only if the copyist has to make a number of key intellectual/creative decisions during the copying process, is a new copyright likely to come about. Obviously I can't comment on whether that might apply in your case, but it seems unlikely on the facts you have given.
As a matter of interest, although it is 'old' law and so would not apply today, one of the earliest court cases concerning photographs was in 1869 and was about an almost identical issue: whether a photograph of a engraving was entitled to copyright. It was known as the Graves case, and you can read about it
here (para 6.1.1).
*
The actual length of the copyright term will depend on a) when the painting was made and b) when publication took place. Suffice it to say, provided that the date of publication was before 1946, copyright will have now expired.