The French patents before 1902, when they start to appear on the free Espacenet database, are very hard to search.

I have traditionally used a variety of tools at the British Library: annual name indexes which do not lead provide a way to the actual publications, granted patents that were never published as the owners failed to pay a fee a year later (the original ought to be at Paris), a subject index in English to 1876 which is arranged by subject title with no cross references — so you have to guess at the heading. The actual publication volumes generally have several to a year, each arranged by subject, and the drawings are at the end with grey outlines…

Recently INPI, the French Patent Office, began putting data on a database called Brevets francais du 19e siecle.  It is all in French. , and coverage is from 1791 to 1855, though I’ve noticed material to 1861. In the results lists, there should always be a “notice” with bibliographic details, and often there is a “dossier” which is a reproduction of a document. This always seems to be handwritten, and is sometimes at least just the formal grant of a patent.

The help notes state that you can put in the “Desposant” field any  of the words madame, madamoiselle or veuve [widow] to find women. Presumably this was noted at the time by the authorities.

There are at present 433 entries for madame, 294 for madamoiselle, and 282 for veuve, making 1009 in all. There are currently 53,528 entries. This suggests 1.8% are by women, a high number for the period. Certainly in the 1850s and 1860s the number of Frenchwomen taking out British patents was high, perhaps even exceeding those taken out by the British (I haven’t fully checked out that period yet). 

However, the field includes “mandataire”, some sort of patent attorney, and there was at least Madame Rabatel with 7 entries in that role. So the actual numbers are somewhat lower.

It is possible to search by profession. Hence a request for “artiste”, run against veuve, gave 5 hits.

My impression is the vast majority of the women, and perhaps the man as well, had French addresses (they could still be say Americans — Robert Fulton has four patents in there, from Paris addresses.

Finally, the help notes strongly recommend using the index which is available, as the little A/B button, to the right of each field (except date) in the advanced mode.

I must admit I found the database a little clunky, but at least it provides a possible way in. It is still a terror to try to get from data found there to any publication in the long series kept at the British Library. There certainly isn’t anything comparable for the early US or UK material. Plans are to eventually go on to 1902.