by Tim Sherratt


  • Trove

Published in Teaching History (Journal of the History Teachers Association of NSW), vol. 53, no. 2 (June 2019).

Ask people to describe Trove in a single word, and they we generally answer ‘newspapers’. And why not? By making more than 200 million Australian newspaper articles from 1803 onwards available online, Trove has changed the practice of history in ways we don’t yet understand.

But as revolutionary as Trove’s newspapers have been, there’s much more to explore. Created by the National Library of Australia, Trove is a digital repository, a collection of collections, and a platform for collaboration and creativity. Not only is it an important starting point for any historical research, it raises critical questions about how digital resources are changing our access to the past.

Not just newspapers

Before we move on from the newspapers, it’s worth reflecting on sheer variety of titles that have been digitised and made easily accessible online. You won’t just find the familiar metropolitan dailies, there’s hundreds of small regional newspapers as well. There’s political papers, religious papers, community papers in French, German, Croatian, and even Chinese. One of my favourites is the Chinese Advertiser (later the English and Chinese Advertiser), which was published on the Ballarat gold fields in the 1850s. Only ten issues are known to survive and they’re all in Trove.

Page of the Chinese Advertiser

The digitisation process creates images of each newspaper page. Then, through a process known as Optical Character Recognition (OCR), it breaks the page up into separate articles, and extracts their textual content. By turning images into text, Trove makes those 200 million articles searchable.

But outside of Trove’s newspaper zone, there are many other digitised resources to explore online. In the journals zone, for instance, you’ll find more than 4,000 issues of The Bulletin from 1880 to 1968. In the late nineteenth century, The Bulletin’s mix of politics and literature fostered both a new sense of radical nationalism, and support for the racist ideal of ‘White Australia’. Other digitised journals include the NSW School Magazine of Literature for our Boys and Girls from 1919 to 1954, the Pacific Islands Monthly from 1930 to 2000, the Australian Woman’s Mirror from 1924 to 1954, the Wireless Weekly from 1922 to 1943, and even the Kennel Control Council’s gazette from 1932 to 1969!

There are more than 300 digitised journals in Trove which, like the newspapers, can be searched at the level of individual articles. However, it’s currently quite difficult to search within a particular journal on Trove. To help with this, I created a simple web app that helps you to explore the digitised journals – just select the titles you’re interested in, insert some keywords, and the app will build your search and redirect you back to the results in Trove.

Other Trove zones – books, pictures, and maps – also contain large volumes of digitised material that you can view and use online. Amongst the books you’ll find a fascinating collection of ephemera – things like leaflets, posters, pamphlets, certificates, and even playbills. There are, for example, a number of items relating to wartime recruiting and the WWI conscription debate.

Conscription poster

The map zone is particularly rich with freely available, digitised content. I did a bit of poking around recently and worked out that there are more than 20,000 maps available online, and about 80% of these are free of copyright restrictions. Like other digitised resources, the maps cannot only be viewed online, you can download high-resolution copies for printing or sharing. In the case of textual resources, like books and journals, you can download complete issues, individual page images, or the raw text extracted by OCR processing – just look for the download tab in the digital resource viewer.

However, as with the journals, finding all of this exciting digitised content can be difficult. There’s just so much stuff in Trove! One simple trick is to include the term “nla.obj” in your search. Digitised resources are given a unique identifier that includes this string, so you can use it to filter your results. If you’re interested in getting deep into the collection data, have a look at my GLAM Workbench. There you’ll find sections on Trove’s books, journals, and maps that include lists of resources and well as bulk collections of full text. Yes, that includes 6.6gb of text extracted from 27,426 journal issues!

We tend to use terms like ‘digital’ and ‘digitised’ interchangeably, but a large proportion of Trove’s digital resources didn’t go through any sort of digitisation process – they were just born that way! ‘Born digital’ works include a growing number of recent journals and books that were published electronically and are available online in formats like PDF or MOBI. But by far the biggest collection of born digital resources can be found in Trove’s archived websites zone. There’ll you’ll find an astonishing 5 billion snapshots of Australian websites since 1996.

But why would historians be interested in a collection of websites? Hmmm, perhaps we should reverse that question. How will future historians understand the 1990s or beyond without using web archives? Collections like these capture a wide range of social, political, and cultural activity that won’t have been documented anywhere else. As we know, websites change and disappear all the time, whether by design or accident. Web archives capture these changes, enabling us to look back in time, treating the internet itself as a historical source.

Seeing things differently

Trove’s web archive raises another important question, just how do you make sense of collections on this sort of scale? What does it mean when you search for something in Trove’s newspapers and discover there are two million matching results? Digitisation has unlocked resources, enabling new forms of discovery and use, but with improved access has come the challenge of abundance. Just what does it all mean?

Fortunately, digital collections also offer new ways of seeing. Instead of reading through a list of two million search results, we can visualise the data in different forms to reveal patterns and changes. QueryPic is a tool I created to visualise searches in Trove’s newspapers. It shows the number of matching results per year, enabling you to examine change over time. For example, using QueryPic you can explore the question, when did the ‘Great War’ become ‘World War I’, by plotting how frequently the two phrases occurred in Australian newspapers. It’s easy to create your own QueryPics – just type in some keywords and press the button!

Conscription poster

One of my recent experiments shows another way of seeing a set of search results. After creating thumbnails from each of the newspaper articles in a search, I pasted them together into one, very big, deep zoomable image. This is what the ‘White Australia Policy’ looks like in Trove’s newspapers.

Conscription poster

There’s a shift here from working with individual historical sources to analysing whole collections. This means extending our historical skill-set to include a range of new digital tools and methods. Once again, my GLAM Workbench provides a wide range of examples if you’re interested in exploring the possibilities further.

But while digital resources like Trove challenge traditional research methods, they also need critical evaluation. As with any historical source, we have to ask questions about context, reliability, and completeness.

Conscription poster

Here’s a chart that simply shows the number of digitised newspaper articles in Trove per year. What do you think caused the peak around 1914-15? Did something significant happen? Yes, it’s because of the war, but not in the way you might think. The peak is a result of digitisation priorities – in the lead up to the centenary of World War I it was decided to focus on digitising newspapers from the war period. If digitisation continues (and that’s dependent on continued funding), the peak will disappear. But for now it’s a useful reminder that Trove itself is a historical construction. While we enjoy the wonderfully rich content it delivers to our browsers, we have to think about what’s not there, what we can’t find, and why.

Trove challenges us in other ways as well. Instead of simply using resources, we can build them. By adding tags or comments to records, by creating lists, or by correcting OCR results in newspapers, we can help other Trove users find what they’re looking for. Through these sorts of contributions we are all building Trove together.

But we can also build completely new things based on the data Trove provides. Perhaps you’d like to play Headline Roulette, a simple game that tests your historical knowledge by asking you to guess the publication date of a randomly selected newspaper article. Even better, you could use the ‘do it yourself’ version of Headline Roulette to create your own customised game – limiting the newspaper articles to a particular topic or region. What about turning a Trove list into an online exhibition, or building a Twitter bot to share Trove collections?

Trove is many things. Most of all it’s an opportunity to connect with Australia’s past in new and interesting ways.

Selected resources

Using Trove

Trove as data

Build and play