Compiling the Horse Register

A Register, Not a Study

Horse Oil Paintings is a register of pictures, compiled rather than researched. Each entry holds a catalogue line — hand, year, medium, size, the collection that owns it — copied from public collection data with its origin named, and beneath it a passage of commentary drafted by a language model from those same fields.

The two are not blended and the boundary is printed. A canvas size out of a museum's database is worth what a museum's database is worth. A paragraph on why a hunting scene is composed the way it is has no author and should be read as such.

What Each Part Rests On

The catalogue line

Open museum APIs, Wikidata, and the Getty Union List of Artist Names for the painter. A field checked against one of them shows that reference in place, numbered. A field not yet checked stands as collected, with nothing implied — the common case here. An entry with no Sources list has been through no check at all.

The commentary

Generated from the catalogue line by a language model. Not sporting-art scholarship, not connoisseurship, and not written by anyone with a name to defend. Use it to get oriented and go to the references for anything else.

The plate

Public-domain reproductions carrying the holding institution's attribution. Terms sit on the copyright notice.

Breed, Harness and What We Will Not Guess

What changes the shape of a picture with a horse in it is the work the animal is doing. A man raised above eye level on a stationary charger is an equestrian portrait, and the subject is the man. A cavalry charge is a battle piece where the horses happen to be the fastest thing in the frame. A plough team is a picture about ground. Racing and hunting are their own trade, with their own patrons and their own painters. Confusing the first two is the standard error in this subject, and titles do it regularly.

Breed is a harder line and we do not cross it. Stud books, breed societies and the specialist literature on English sporting painting are where a horse gets identified as a particular type or a particular animal; none of that is in the records we read, and reading conformation off a canvas is not something a data pipeline can do honestly. Where a museum names the horse or its owner, the entry repeats it. Where it does not, the entry says nothing, and silence here is not modesty about a fact we hold back — it is the absence of the fact.

Assembly is also unusual on this site, and worth admitting. Barely a third of the pictures that belong here say so in the title — The Ploughman's Rest has a horse in it and names none — so inclusion leans on the subject tag in the source data as much as on the title, which means a person's judgement about the picture is doing work that no pattern could do — and a person's judgement is exactly what a pattern cannot audit. Works are missing. If one is missing that you know of, say so.

What This Register Is Not

Claims routinely made by reference sites, all of which would be untrue here:

• Nothing published here has been refereed. There is no review before publication of any kind.
• No board of specialists, no consulting equestrian historian, no member of staff holding an advanced degree.
• No owner, trainer, curator or dealer has been interviewed for any entry.
• No unpublished archive is opened, and no stud book, sale catalogue or breed register is consulted before an entry goes up.
• No museum, collection or racing institution partners with this site. We read published open data and nothing more.

It is better to describe a modest operation exactly than to dress it up. Where another page here says otherwise, that page is wrong and this section stands — let us know.

Human Review

No one is published as editor of record. Until someone is, take it that no human has read the commentary on an entry, and rely instead on the catalogue line and whatever sources it names. A masthead without a person behind it is worth less than an admission.

The collection is assembled by Louie Drehmer, who chooses the pictures and runs the pipeline that fills the fields. Choosing and checking are different kinds of work, and one is not offered here as proof of the other.

When Catalogues Differ

A sporting picture may be dated by the race it shows in one catalogue and by its exhibition year in another; a canvas may be measured before and after a nineteenth-century reline; a work may change owners faster than any database updates. None of that is settled here. Sources are queried in a fixed sequence, the first value found for a field is kept, and rival values are discarded rather than recorded.

The entry does show which record its value came from, named and linked, so the disagreement is yours to inspect. That is a narrower promise than resolution, and it is one the software keeps every time.

Attribution passes through untouched. Where a record already reads "studio of" or "attributed to", so does the entry; we never add such wording ourselves, and its absence should not be read as confidence.

Corrections

Mail post@horsespaintings.com with the entry URL and the error. A catalogue fact is checked back against the source record, and the entry then displays the day its facts were last verified, printed on the page rather than logged out of sight.

Any change to the substance of an entry moves that date. Where the records will not decide a question, the entry keeps the doubt visible instead of resolving it silently.