What Google can’t find

The California State Library holds millions of images documenting the state’s history, but only a small fraction can be found online. Its collections range from daguerreotypes, negatives and prints to postcards, photo albums and early images of places such as San Francisco and Yosemite. As Newsjunkie expands its archive directory to include image collections, managing editor Peter Landau spoke with Jessica Knox-Jensen, who runs the California State Library Services Bureau, about how these collections are built, how researchers can uncover material that a Google search will never find, the gaps and biases embedded in the historical record, and why provenance is becoming even more important in the age of AI.

What exactly is the California State Library Image Collection? What makes it different from a collection of photographs?

An archive in general is a group of materials preserved because of their enduring value. Images are a way to capture a moment in time, and the California State Library has a really extensive collection of images in all formats: slides, prints, negatives, digital images and special formats like daguerreotypes, tintypes, ambrotypes and orotones. We have photo albums, mug books, postcards, lithographs and other visual materials featuring images.

The real point to drive home is that while there are some materials online, that is just the tip of the iceberg when it comes to our archive of images.

When somebody walks into the archive, as opposed to logging onto your site, what are they most often looking for? What do they end up discovering that they didn't know they were looking for?

Lately there has been a lot of interest in the diverse people of California and traditionally marginalized populations, although what we have is often captured from an Anglo-centric viewpoint. There's also a lot of interest in 19th-century images in all formats—people, the built environment and landscapes.

It's difficult because archival collections are not traditionally organized by subject, which is why cataloging and indexing are so important to facilitating public access. We've traditionally cataloged images at the item level.

There are many different ways to access the photo collection, whether you're searching the online catalog for catalog records and digitized materials or searching our physical card catalog, which is still one of the main access points for our photo collection.

You still have a physical card catalog. That's unusual, isn't it?

We do. It's really a function of the size of the collection. We have millions of images, and all of the information contained in the card catalog has not been transferred to the online catalog. So we really have two points of access, and we check both when we're doing photographic research for people.

What in the collection would surprise people the most?

What fascinates me most are the images in special formats from the early history of photography. One of my favorite things to show off are the orotones [a vintage photographic print made by exposing a positive image on a glass plate coated with silver gelatin or collodion emulsion—ed.]. We used to say that they were specifically made with a banana-oil emulsion, but that's not necessarily correct.

The processing chemicals have a really fruity smell. From the early 1900s, we have a lot of orotones, which are reflective and gold. We have images of Yosemite Valley. We tried to digitize them and put them in the catalog, but nothing really captures the sparkle that you experience in person. That's one of the really surprising things.

Is there an image or collection that you find yourself showing people again and again?

Definitely those orotones of Yosemite. Those are really important.

There's another collection that we get a lot of requests for, and that is the Stillman collection. Those are photographs of early San Francisco Chinatown.

We also have a lot of photograph albums and photographs of early Californians, including people of political significance. We have several iconic images from the women's suffrage movement as well.

What does it take for an image to make its way into the State Library, and who decides what is worth preserving?

We take images in all formats. We've acquired digital images from donors, and we still take physical images. We can receive images one at a time or as part of a larger collection donation.

We have a collection development policy on our website that articulates the goals we have for the collection and what the different section heads consider when they're evaluating materials for acquisition.

Typically, we want to tell the story of all Californians. We want to make sure our collection reflects the diverse history of the state, so we're looking for visual materials that tell that story and allow people to find themselves represented in the State Library's collection.

We try to meet with donors, and we always do deeds of gift. We've taken digital materials via email, hard drives and USB drives. We take physical materials that have been mailed to us or dropped off. There are many different ways things can find their way into the collection.

When you're collecting, are you thinking about future historians, preserving what survives from the past, or both?

In some cases, it's both. It's difficult to anticipate what the research needs will be in the future, so we want to build as comprehensive a picture as possible rather than focusing on just one area of scholarship.

However, current research trends definitely inform what we're bringing in at the moment. It's a difficult balance between selecting materials that meet the needs of current researchers and what will meet the needs of people in the future.

You mentioned that not everything is searchable online. How much is, and what does a researcher miss if they assume that what comes up in a digital search is all there is?

At the moment, I believe we have somewhere in the neighborhood of 36,000 images in our online catalog. But we have millions of images in the stacks.

That includes more than 500,000 emulsion-based negatives, 55,000 slides, 230,000 prints, more than 300 photo albums, over 40,000 postcards, and 250 daguerreotypes and other cased images.

The online search is just the tip of the iceberg.

What can an archivist find that Google, or even a very good archive search engine, wouldn't provide?

A Google search will pull up a lot of images, but it isn't comprehensive. Google cannot typically search within library catalogs and databases.

Those are usually systems that are separate from a repository's webpage, which is what Google picks up. That limits what you can actually discover in a Google search, which is why it's important to identify different repositories and search their individual catalogs so you can connect with the resources they have available.

I also think we are a credible source of information. Repositories like ours strive to be transparent about where our material originated and what we know about it. We also aim to preserve materials for future generations.

Citing our collections as a source is more reliable over time than linking to an image from a source you haven't vetted that simply came up in a search.

I would encourage people to contact repositories and work with research librarians and archivists because they know the depth of the collection, and that often can't be captured through an online search.

When somebody comes to you with a vague request, such as, "I need pictures of Los Angeles in the '30s," how do you turn that into a productive archive search?

We go through the reference interview process. We want to find out as much as we can about why somebody is looking for something specific, because that might give us other facets to search.

We'll try the online catalog, the photo card file and the postcard file to identify relevant subject headings.

In the online catalog, those subject headings are searchable, so you can click them and pull up digitized images related to that specific subject. Often, the subject heading assigned to something might be different from the keywords that your online search would find. Using that methodology can pull up more material of interest than simply searching with the terms you're already familiar with.

We typically work with everyone one-on-one to refine the request and find materials relevant to their needs.

What should a journalist or researcher provide to improve the odds of finding something useful?

The potential use for the materials is important. What you're trying to accomplish by finding that material can help us narrow down the different categories that might be useful to you.

Perhaps you're looking at a time period where the photographic record is a little thin, but you might be able to find sheet music with illustrations, lithographs or other material that would meet your needs even if we don't have photographs.

The other important thing is being open to a referral to another repository. We often have overlapping missions, and another repository might have exactly what you're looking for. Because we're aware of that, we can connect you to those resources as well.

How do you know what you're looking at? An old photograph may have no caption, an incorrect date or unidentified people. How much detective work goes into describing an image accurately?

It's an extensive amount of work. There is both the image-identification component and the cataloging component.

We have subject-matter experts on staff who are familiar with the different tells for a particular time period—what different formats indicate about when something was produced and what technology was used. There's also checking other library catalogs to see whether there is related material that might look similar.

Being familiar with California's landscapes, architecture and landmarks can help you identify a particular place or person. So it is a lot of detective work.

We also pull a lot from context. In many cases, there are things written on the backs of photographs. We've also had things written on the front. All of our staff who work with historic materials are really good at interpreting 19th-century handwriting, which can be difficult if you're not familiar with it.

There's an extensive amount of work that goes into identification. We catalog at the item level, which is very time-intensive for staff. That's why the majority of our material is uncataloged—and why what you do find cataloged is incredibly detailed.

What happens when the description attached to an image reflects the assumptions, terminology or biases of the period in which it was cataloged? Do you rewrite those descriptions, preserve them, or both?

We have a harmful-language statement on our website, and we're actively engaged in reparative cataloging.

There's a big difference between transcribing the content of a caption on a photograph and the descriptive information that generations past may have used to identify it.

While we aim to preserve the original text or language, however offensive it might be, because that is of value to historians, we're actively engaged in rewriting descriptions applied by library professionals so they don't use harmful language reflective of past practices.

Whose California is well represented in your archives, and whose California is harder to see?

We've collected strongly in areas that were of interest to historical research 50 to 100 years ago.

We have a lot of photographs documenting major political figures and people of political significance throughout the state since the advent of statehood. We have fewer photographs documenting the experiences of women and marginalized populations.

The photographs we do have documenting Chinese Americans or Native Americans certainly have an Anglo-centric perspective to them. We're looking to expand the collection in that regard and document the experiences of those who may not have had a voice in the past.

Of course, that is also limited by what photographs or images exist.

Are there communities or periods where you know the historical record has major gaps?

Yes, absolutely. Generally, those would be people who were not part of the mainstream narrative.

That includes communities such as Japanese Americans, Chinese Americans, Native Americans and women, as well as people who may have dissented from prevailing political narratives at the time and people involved in the labor movement.

We would love to add more and ensure the historical record is comprehensive and balanced.

It also means revisiting the collections we already have to make them more discoverable. For example, our Sutro Library kicked off something called the Misses Project. The majority of the images of California women of political or historical significance were cataloged under their husbands' names.

They spearheaded an initiative to do the research, give those women their names back and catalog the images under the correct names rather than their husbands' names, so somebody searching for them would be able to find them.

For journalists in particular, what are the most common mistakes people make when using historical images?

Taking them out of context or not providing the historical context can be damaging and harmful in some cases.

That takes me back to the harmful-language statement on our website. Where there are indicators of misogyny and racism, we want to provide context rather than simply plucking something out and including it with a story. We want to provide more information to help people understand the images and recognize that the values reflected in those primary sources may not reflect the values we have today.

I also think it's important to do your own research, because everybody makes mistakes. Sometimes donors make mistakes, too. Family lore doesn't necessarily align with the historical record.

If something seems off, it's certainly better to question it and do some additional research than simply go with it.

There could be a misidentification.

Right.

The date could be wrong.

Absolutely.

The image could even be cropped in a way that changes its context. Those are all things a researcher needs to take into account. Finding an image and having the right to publish it are also two different things. What should a journalist understand about permissions, copyright and reproduction before using something from the archives?

We strive primarily to digitize materials that are in the public domain, but we don't make interpretations in terms of copyright unless the copyright has been specifically assigned to the library.

The burden of clearing copyright falls entirely on the user.

We can provide permission to use digitized images or images from the collection based on the fact that we own the physical materials, but we can't clear copyright.

Wherever possible, if there are copyright restrictions and we're aware of them, we provide that information. In some cases, we may decline to make a reproduction.

Have you ever personally opened a box or examined an image and realized you were looking at something historically important that had essentially been forgotten?

Yes, absolutely. Prior to being an administrator, I worked for the State Archives for 10 years, and I have worked with collections here.

I can't recall one specific example, but it happens every day.

Recently, our subject-matter experts in California history who work specifically with visual materials have undertaken a massive project to digitize our nitrate and emulsion-based negatives.

They're currently going through half a million images to quality-check them and make sure we were able to salvage some of them. What they've found is really incredible. There are images of San Francisco right after the 1906 earthquake and fire.

Those kinds of discoveries happen every day, and that's why we work so hard to share them online.

Is there an image in the collection whose meaning changed once you learned the story behind it?

I think we all look at things differently now that there is an effort to reframe history in a broader context, to be more inclusive and recognize the struggles that others may have endured.

We can look at images of figureheads who were once revered in California history and now understand a bigger picture—where they may have done things that would not align with our current values and that damaged people or had horrible impacts on other populations.

In general, that kind of shift has been really important.

May I ask the strangest request you've received?

With the caveat that I don't typically handle the day-to-day requests—our subject-matter experts do a lot of that—our social media team often goes through the catalog to find images that are on point with current social media trends.

It's not the strangest request, but one thing that continues to resonate with me is how people have always been cat and dog people. We have photographs of people who dressed up their cats that we share for National Cat Day.

One of the things I take away from managing the collection overall is that people don't change that much over time.

Cat person or dog person?

I'm a cat person.

Me too. One final question: What image in the collection would you save first if you could save only one?

That's very difficult. There is an image that I think is really important to the history of the state. It's a daguerreotype of African American miners in the gold fields in California.

In fact, I believe there are two of them. They're amazing survivors because of the format and the time period. They're also one of those rare instances where we have visual evidence of a population that didn't receive as much attention in the historical narrative up to this point.

It's so important. I can certainly share a link to that.

I lied—I have one more question. AI can now generate convincing historical-looking photographs and search and describe enormous image collections. Does that make the role of the archive and the provenance of an image more important?

Yes, absolutely. I think this is a challenge we will be navigating for quite some time.

In terms of donations, being able to go back to the original or know whether it has been reformatted, where and when, is the best thing you can do to preserve the historical record.

There are still a lot of problems with AI creating hallucinations or misidentifying things. As we continue to refine that technology, I think it's really important that libraries advocate from an informed position for the responsible use of AI in libraries. If we do employ those tools, they should augment rather than replace the functions we already perform.

When we're talking about authenticity, it's definitely a major challenge for the archives and library communities. I think it really underscores the importance of information literacy and the role librarians and archivists play in teaching people how to evaluate resources for credibility.