A maker’s name attached to an old dolls house miniature can make a considerable difference to the way we view it. An anonymous little chair suddenly becomes a Barton chair, a tiny metal accessory is described as Gerlach, or a small wooden cabinet is attributed to one of the recognised British manufacturers. That identification can influence the history we attach to a piece, how desirable it is to collectors and, inevitably, its value. Yet there is something we perhaps do not talk about often enough in dolls  house collecting: an attribution is not always a fact.

So can you trust a dolls house miniature attribution? Sometimes absolutely. Sometimes probably. Sometimes there is simply not enough evidence to know. And do people get attributions wrong? Yes. They do. The most knowledgeable researchers, collectors, dealers and authors have all made mistakes, and anyone who has spent long enough trying to identify old miniatures will almost certainly have changed their mind about something at some point. I certainly have. That is not necessarily because the original research was careless. Very often it is because another piece of evidence has appeared.

We have to remember what these objects originally were. Most were toys, not historical artefacts carefully preserved for future researchers. Packaging was discarded, labels fell off, children repainted furniture, broken handles were replaced and sets were separated. Dolls houses themselves could remain in a family for decades, gathering furniture and accessories from different periods as they passed from one child to another. By the time an individual piece reaches us today, virtually everything that once told us where it came from may have disappeared.

Even manufacturers who labelled their products did not necessarily mark every item they produced. A suite of furniture might contain several pieces but only some retain labels. Tiny accessories might originally have been identified by packaging rather than by any mark on the object itself. Once that packaging disappeared, so did the easiest means of identification. Much of dolls house research therefore involves trying to reconstruct an object's history from the evidence that survives.

Not all evidence is equal. An original manufacturer’s mark or securely attached label is clearly very important, particularly when everything else about the piece is consistent with what is known of that manufacturer. Contemporary trade catalogues can be even more revealing. Finding the same design illustrated in a manufacturer's catalogue, perhaps accompanied by a catalogue number, can provide very strong evidence indeed. Occasionally we are fortunate enough to have both a labelled example and a catalogue illustration, and at that point the attribution becomes particularly convincing.

More often, however, we are working somewhere between certainty and possibility. An unlabelled chair might be identical in dimensions, construction, finish and upholstery to a labelled example. A cabinet might use the same handles, feet and construction techniques seen repeatedly on documented furniture from a particular manufacturer. Those are all useful clues and together they may form a very persuasive case, but the further we move away from direct documentary evidence, the more careful we need to be about the language we use.

Looking alike is not always enough. Anyone who handles large numbers of old miniatures begins to recognise certain shapes and construction methods. A particular chair back feels familiar, a handle resembles those seen on another range, or a painted finish immediately reminds us of a known manufacturer. Experience is enormously valuable, but visual recognition can also lead us astray. Manufacturers were working in the same period, responding to the same fashions and producing for the same market. Two companies could make very similar dining tables, wardrobes or chairs simply because both were reproducing furniture styles that were fashionable at the time. The same problem occurs with accessories. A tiny candlestick, vase or kitchen utensil may strongly resemble a documented example without necessarily having come from the same factory.

This is why a good attribution is usually built from several clues rather than one. Construction, materials, dimensions, fittings, finish, decoration, provenance and documentary evidence all contribute to the picture. One characteristic may suggest a maker; several independent characteristics pointing in the same direction make the attribution much stronger.

There can sometimes be an assumption that if an attribution has been made by a respected researcher or has appeared in a specialist book for many years, it must be correct. Unfortunately, research does not work like that. The most knowledgeable people in this field can and do make mistakes. Someone may have collected for fifty years, written extensively on a manufacturer and studied hundreds of examples, yet still identify an occasional piece incorrectly. Expertise does not mean never being wrong.

In fact, some of the most important early work on dolls house manufacturers was undertaken under circumstances very different from those available to researchers today. There were no searchable online archives, digitised newspapers, enormous photographic databases or instant access to collections around the world. Researchers often worked from the catalogues they could physically locate, objects they had personally handled and information exchanged between relatively small networks of collectors. Their work remains enormously valuable, but it should not be treated as incapable of revision. A previously unknown catalogue may surface, an original box may appear with its contents intact, or a securely labelled example may emerge that proves an attribution accepted for years was incorrect. The earlier researcher may have reached an entirely reasonable conclusion from the evidence available at the time. The evidence has changed, and therefore our conclusion should be allowed to change with it.

Repetition presents another difficulty. An attribution can acquire an extraordinary degree of authority simply by being repeated. An unmarked miniature may appear in an auction catalogue attributed to a particular manufacturer. A dealer buys it and repeats the description. Another collector sees the dealer's listing and uses it to identify their own example. Photographs begin appearing online carrying the same manufacturer's name and before long a search produces twenty apparently independent examples with exactly the same attribution.

Twenty examples sound convincing, but if every one of those identifications ultimately traces back to the same original assumption, we still have only one piece of evidence.

This is particularly important now that so much collecting research takes place online. Search engines are extraordinarily useful, but they can also create an illusion of confirmation. Seeing the same attribution repeatedly does not tell us whether ten different researchers independently reached the same conclusion or whether nine people copied the tenth. The important question is not how many times an identification appears, but where it originated and what evidence supports it.

The language used when describing miniatures therefore matters. Saying that something is by a manufacturer suggests a high level of confidence. Saying it is attributed to a manufacturer acknowledges that the evidence points strongly in that direction but may not be definitive. Possibly by indicates greater uncertainty. In the style of describes resemblance and should not be confused with an attribution at all. These distinctions may appear small, but they tell a collector something important about how secure an identification actually is.

Then there are the pieces for which the correct description is simply maker unknown.

There is nothing unsatisfactory about that. In fact, admitting that we do not currently know who made something is far more useful than attaching a well-known manufacturer's name without sufficient evidence. An unidentified miniature has not failed some sort of collecting test. Its history is simply waiting for more evidence.

Marks themselves can also be taken further than the evidence allows. A miniature impressed Germany, for example, gives us valuable information about its country of manufacture, but it does not necessarily tell us which German company produced it. A particular construction technique may support an attribution without proving one. An old collection can establish that a miniature existed by a particular date without identifying who made it. Every clue has limits, and good attribution depends partly upon recognising those limits.

Old books, auction catalogues, museum records, dealers' descriptions and collectors archives should all be treated as sources rather than unquestionable authorities. They can contain extraordinarily important information, but no source is incapable of error. Printed information can be wrong just as information online can be wrong. Equally, something should not be dismissed merely because it comes from a collector rather than an institution. What matters is the evidence behind the claim.

Perhaps the most important part of attribution is being prepared to change it. If new evidence appears tomorrow showing that something we have described for years as one manufacturer was actually made by another, there is no virtue in defending the old identification simply because we have used it for a long time. The responsible response is to correct it.

That is not a failure of research. It is research doing exactly what it should do.

There will almost certainly always be discoveries left to make in dolls house collecting. Catalogues remain hidden in private collections, old packaging survives unnoticed in boxes, family dolls houses occasionally emerge with their original contents and previously undocumented pieces continue to appear. One of those discoveries may confirm something collectors have suspected for decades. Another may prove that something we thought we knew was wrong.

That possibility is part of what makes researching these tiny objects so fascinating.

So can you trust an attribution? Yes, when the evidence justifies that confidence, but the name attached to an object should never be accepted simply because it has been repeated frequently, appears in a book, comes from an auction house or has been supplied by someone regarded as an expert. The better question is always: what evidence supports it?

Expertise does not mean never making a mistake. It means recognising the difference between what is known, what is strongly supported, what is suspected and what remains unknown. Every serious collector and researcher will occasionally get something wrong. What matters is the willingness to reconsider an attribution when better evidence appears.

After all, the purpose of identifying these wonderful little objects is not to prove that we were right. It is to discover as much of their history as we possibly can. And sometimes, after the catalogues have been searched, documented examples compared and every surviving clue examined, the most accurate attribution we can give is still simply maker unknown.