We need to make women in photography collections visible - but how?

I'm currently working with one of S?rmlands museum's fascinating photography collections. It's a portrait collection from Mariefred, a small town in S?rmland, with about 1200 images on glass plates. They've been online in our collections for a long time - but with only minimal metadata and low image resolution.*

As S?rmlands museum has been able to increase and intensify their work with digitisation of our collections, this is one part of a larger collection that we've been working on. The whole collection was donated to the museum by Ernst Lundqvist in the 1970s, comprising almost 2000 images on glass plates. 1200 of them were taken by a pioneer female photographer, Anna Wissler (1857-1920), who had a studio in Mariefred.

I find portrait collections like hers quite fascinating. We can see people in quite similar surroundings, with small details giving away the differences in their cirumstances. We see a lot of men in suits, newly wed couples and children in white dresses - and a couple of dogs as well.

A lot of cultural heritage museums or archives have these kinds of collections, but one of the difficulties in digitising and making them accessible is actually how to make them relevant, both for the institution and for the user. Museums rarely use them in exhibitions and a majority of them will, when digitised at all, be published quite anonymously, with only rudimentary metadata. I recently participated in a workshop at J?nk?pings l?ns museum where these database posts were called ghosts, a term that fits even better to these photograph collections with thousand of people staying ghosts instead of becoming identified people.

With Anna Wissler's collection, we've had luck. A majority of the photographs had notes on them, with a name, a date, and sometimes even a place. However, the name on the image does not necessarily match the person on the image; it might just be the one ordering the portrait, a parent, or, in the case of many women, the husband's.

Although these images have now been available online, we know that there is only a small number of them ever used. When we re-digitised them last year, scanning them was only one part of the work. Another one, which is still ongoing, is adding more detailed metadata. And that is the much more challenging part.

It's not only a task that takes a lot of time. It is also based on access to data from other institutions and sources. Compared to other countries, Sweden has a lot of genealogical research data accessible online, partly in open, public databases such as the Riksarkivet Swedish National Archives ' ones, partly via paid portals such as Ancestry. Furthermore, this is a typical research task that needs a thorough definition of your individual level of "good enough".

Let's take a step back and look at why it is in our interest to add as much information as is possible and necessary to these digital images. Here's an example:

Because of her unfortunately very common name, it was not possible to identify Agnes Andersson beyond the name attached to the image with the accessible databases, even when trying to narrow down her birth date. However, this is one of the many images in portrait collections that have so much to tell than the actual subject of the image. In 1901, she is one of very few women choosing (or being told to) be portrayed in a folk costume; most wear contemporary dresses or a skirt and a blouse. The costume in question isn't even a real folk costume. It has parts from the R?ttvik folk costume, but most of it is assembled and imagined - making it more a costume than a folk costume. It probably was a part of Anna Wissler's props equipment. Looking at the date when the image was taken, it tells a lot more about Swedish romantic nationalism and the longing for identity and folklore - especially when we also look at the next image.

There are cases like this image, where we find the same person being photographed at different times, or as in this case, on the same day, but in different clothes. Linking the images together makes it not only easier for everyone to find for example Agnes Andersson in our collection. By adding context about this particular day, one woman being portrayed in a decorated space with a staged scene and folk costume compared to the rather neutral portrait in a contemporary dark dress, tells us (and the users) much more than looking at the photographs separately.

Managing ambiguity

One of the major challenges in identifying and adding context to the portrayed people is accepting and displaying uncertainty and ambiguity. Here's an example for that:

A decade or two ago, I might not have found any information about this image within the reasonable amount of time I have set myself for my own level of "good enough" research. It would have been one of the images on the digital mountain of images that leave me frustrated because I cannot provide any more information than the original notes. Today, the situation is thankfully enough different. However, we often find an abundance of information that we need to interpret and proritise.

The information I had was "Doktor Thelberg, Nyn?shamn, 19 September 1907". I suspected at first that it was not the person depicted, but maybe a father or husband to the woman depicted. However, I soon found a Doctor Elizabeth Burr Thelberg (1860-1935), a female physician from the US, married to a Swede living in New York. She became famous for her work during the First World War in the American Women's Hospitals in France. It could as well been her daughter, the other Elizabeth Thelberg (1891-1953). From their migration papers available, I could however not find a definite prove of that they had been in Sweden, let alone in Nyn?shamn, in 1907.

A wider search for Thelberg and Nyn?shamn pushed me then in the presumably right direction. The woman we see in the picture is probably Eva Augusta Thelberg, née Duryea. Her biography illustrates why we never content ourselves with the names of husbands, fathers, brothers, but give these women their own names back.

She was the eldest daughter of Edgar and Julia Duryea, born in 1866 in Queens, New York. She was a member of a wealthy family on Long Island, her grandfather being Hendrick Vanderbilt Duryea, the founder of a starch factory in Glen Cove.

On 9 September 1893, she married Martin Ansgarius Hummerhielm Thelberg (b. 1860 in S?br?, V?sternorrland). He had emigrated to New York, USA in 1885 after studying pharmacy. He trained as a doctor at Bellevue Hospital Medical College. From 1898 to 1902 he worked as a physician and surgeon in New York. He was actually the half-brother of the Swedish physician married to Elizabeth Burr Thelberg, my first trace in this research.

Eva and Martin Thelberg emigrated to Sweden in the early 1900s, but they returned to the US several times, including in 1911 when her brother Walter E. Duryea died (a person with a whole interesting story himself), in 1922 and in 1929. Martin Thelberg worked as a doctor in Mariefred (the place where our photographer Anna Wissler had her studio) and Nyn?shamn (the place registered in the order). Eva Thelberg inherited a considerable sum from her family, making her indepently from her husband wealthy - which also would explain her dress, standing out from most of in the collection. After his death in 1937, she donated an antependium of black silk damask with silver embroidery in his memory to Sorunda church.

Being open about uncertainties

However, in our publicly available database, I leave room for ambiguity in the image description. Although I have in the end gathered enough information to regard it as relevant and viable enough to include it, there is always the possibility that she was just a guest of the Thelberg family, giving their address for the order. Is it then responsible to publish this information at all?

With many objects and photographs cultural heritage collections, if not the majority, there is always a point where we can make educated guesses and have to rely on professional expertise and experience. If we don't write those down and publish them (as always, with the right amount of context and sources), our colleagues following in our footsteps in the database or our audience online will never have the chance to verify or correct our assumptions. Maybe the person still has family out there? Writing down a name that we today cannot verify to 100 percent increases the chance that another person googling it to find this image in S?rmlands museum's collections. In the best case, they will find my contextual information helpful; otherwise, they might be able to help us to make it better.

Can we do this kind of research for every image in our collection? No, not in this depth, and not even for each and everyone. However, we can prioritise some groups that have been marginalised before. Some are low-hanging fruits when it comes to adding women's names, such as images of married couples, identified with just the husband's name (as long as it's not that usual in your region [looking at you, Anders Larsson]), even better when there's a place or date attached. Sometimes, clothes give away a lot about the person, that can help you find them through their occupation (or at least help you identify them in a considerable amount of people with the same name). Trying to narrow the search down with guessing the age of the person and thus their year of birth can also be helpful.

Some of this work is quite tiresome. I do this image by image, looking a couple of seconds at each image and its metadata with two things in mind:

1. Do I see a chance to actually identify this person?
2. Is that necessary? If their name is already registered, is that enough? (In many cases, it needs to be) Or do I see something in this image that leads me to think that there is more to tell about their circumstances?
3. If there's no chance to identify them, is there any chance to make this image better to find and work with? Then I try to add information within the defined level of a "good enough" image description.

What are the conditions of being able to work this way? A good digital copy, allowing for zooming in on details, a group of colleagues helping you when you're out of your depth (hello, folk costumes), access to digital databases and a bit of experience in searching for information in them.

Is that the future though?

Let's be honest, this type of work can be incredibly rewarding when finding people like Eva Thelberg (and so many more). However, a lot of times, you just have to give in and accept that you'll never find this Per Larsson's wife. I am int the incredibly luxurious position that I am able to do this kind of work in elevating the quality of our (digital) collections, but a lot of institutions do not have resources to do this.

There are ways to outsource or enhance these types of tasks. Some have worked really successfully with crowd-sourcing image tagging or image descriptions, for example on Wikimedia Commons (see the report from the Wikimedia Commons Data Roundtripping Project) or similar crowdsourcing platforms (see Arolsen Archives' project "Every name counts" on transcribing documents from concentration camps). Then there are others working on and with AI tools to identify things in and across their collections (see a Dutch project about species, the Science Museum Group's Congruence Engine project, and, for context, a recent study on "AI and?Swedish Heritage Organisations: challenges and?opportunities" by Anna Foka, Elisabeth Wennerstr?m & Gabriele Griffin). However, right now, we are as a sector in much more of a prototype-and-test phase - and the majority of us is not even there yet. Most institutions do not only lack resources to adapt similar tools, but even skills to make informed choices in evaluating the results of such tools.

These are just some thoughts gathered after a lot of genealogical rabbit holes, and I'd love to know more about how you are working with similar collection, in the comments or via DM.

* Why did I write this piece? I kind of miss the exchange formerly provided by Twitter, and I felt the need to write some ideas and workflows down to discuss them with colleagues. This is far from any kind of new thoughts or so-called innovation, and much more of a search for alternative online ways to talk about the very common stuff we do at work.