A lock of hair, flowers, inscribed napkins, audio tapes of various standards, letters from Nazi imprisonment – and, of course, many other paper documents – all this is contained in the estate of the prominent labour lawyer and social democrat Hugo Sinzheimer.
Estates document the actions, motives, thoughts and personal environment of the deceased. This provides an individualised perspective on political, cultural and social developments. These perspectives can reveal areas of life that cannot be captured by the records of organisations and are particularly diverse and valuable in the case of Hugo Sinzheimer.
Who was Hugo Sinzheimer?
Hugo Sinzheimer was born in Worms in 1875 and was a lawyer, SPD politician and honorary professor of labour law during his lifetime. Through his work in these various fields, he succeeded in helping to establish German labour law and, at the same time, securing its status as an independent field of law. His Jewish religion and membership of the SPD made him a target of the National Socialists in two ways. Despite being arrested several times in Germany and later in the occupied Netherlands, he and his wife managed to survive the Nazi era. Sinzheimer's estate therefore offers a unique perspective on German-Jewish (legal) history in the first half of the 20th century.
The AdsD hybrid scanning station
In order to make the estate as widely accessible as possible, we decided to digitise it. We used a hybrid scanning station as our digitisation tool. It consists of three components:
1) An overhead scanner for fragile objects up to DIN A2 size: Here, the objects are placed on a padded book cradle and secured with a fold-down glass plate. A scan is then made using a camera permanently installed above the glass plate.
2) A feed scanner, specially modified for cultural artefacts in good condition: Here, the objects are transported into the scanner on tapes, scanned and then ejected again.
3) A computer that connects both scanners and can be used to control the scanning software.
The simultaneous control of both scanners makes the scanning station a ‘hybrid’ scanning station: this allows to decide which scanner to use for which object, down to the individual sheet level. The digital images generated by the two scanners are merged directly.
Digitisation and transcription
During a brief initial review of the objects, we performed a condition assessment and recorded the number of objects. We then determined the appropriate scanners for each of the objects.
The next step was to extract the texts from the digitised images. The technology we used for this purpose is called Handwritten Text Recognition (HTR) and is offered by the Transkribus platform, among others. Essentially, the scans of the objects are automatically broken down into lines of text and deciphered using artificial intelligence.
However, we encountered a problem during the actual scanning process: although Hugo and Paula Sinzheimer's letters were all in a condition suitable for the feed scanner, most of them were in two columns. Unfortunately, in initial tests, Transkribus was unable to recognise the layout correctly and sort the lines in the correct order.
We therefore decided that the multi-column letters should be digitised using the overhead scanner rather than the feed scanner, as this allowed us to separate the pages manually. Although this slowed down the digitisation process considerably, it meant that the digitised letters could be split up, i.e. one image per column. This was also important to us because we wanted to keep open the possibility for our users to analyse the correspondence using artificial intelligence in the future.
It took more time to scan the letters, but we needed significantly less time than originally planned to scan some of the bound notebooks. The processing of the two analogue audio recordings contained in the estate was also completed quickly as part of a larger-scale digitisation project commissioned by the AdsD. By using our feed scanner for single-column letters, we were able to speed up the digitisation of the estate enormously without damaging the objects.
Difficulties with AI-assisted transcription
All digitised documents from the estate were processed using Transkribus. Although the result was perfectly usable, it did not achieve the quality of a professional transcription. Despite errors, it must be emphasised that the text created by Transkribus greatly eases the palaeographic entry into the various manuscripts of the Sinzheimer family.
The end result of this project is the general availability and searchability of approximately 4,700 fascinating pages of German-Jewish (legal) history from the first half of the 20th century, comprising letters, notebooks, photo albums and individual photos, newspaper articles, certificates and passports. And to answer our question from the beginning: we digitised the lock of hair, very carefully, using the overhead scanner.
The digitised estate is available online here.