The Reconstructed Patent Registry of Mandate Palestine Search FAQ

The search options are meant to be user-friendly and intuitive, and where it is not, please blame the British Government of Mandate Palestine!  We invite you to browse, explore, play, use, and share! General Queries

Yes!  If you wish to explore the raw data – you are more than welcome to do so. The underlying database is available for download here.

We suggest:

Michael Birnhack, Mandate Palestine’s Reconstructed Patent Registry (1924-1948) (Tel Aviv University, 2025), available at https://en-law.tau.ac.il/MandatePalestineIP.

• If you want to browse the dataset, the best way to see all applications is to search for “application” – which will provide all applications in the dataset and then apply further searches.
• By knowing the application number.  This is a good start. Be careful with deducing conclusions about priority from the numbers. Note the comments in the Methodology section and the Background section about the various channels of registration, which means that a number is only a good enough proxy for the date, but not a perfect one.
• By knowing the applicant’s name.  This is also a good start.  Note that names may not be consistent: A person or company’s name may appear in several versions. We grouped them together where there was no doubt, e.g., “Flower Inc.” and “Flower Incorporated,” but where the names changed, we left them in the original form.
• Some countries had only a few applications, so if this is what you are interested in, searching for these countries may be more effective than specific searches.

Yes!  You can search for patents according to the product they describe.  You can do this search by two methods:

• • If you know the patent’s field, you can search or filter your search results by the patent's categories. Note: the categorization was not part of the original Mandate data and is our making.
  • If you don't know the technological field, search "kind of product / services." 

Yes, you can search the database using keywords.  We have applied Optical Character Recognition (OCR) technology to scanned PDF files to make the text machine-readable, enabling keyword searches in patent specifications and claims.  However, please note that due to the age of the documents and some handwriting applied to them, the OCR may not be accurate. OCR technology does not always recognize handwriting or manual erasures, both of which were common in these historical files. Users should cross-reference the OCR results with the original scans for verification.

You can search for patents in a specific language or a combination of languages. Submit an initial general query (“application” will provide the entire dataset), and then filter (on the right hand) according to the languages.

To begin with, these AI-generated summaries provide a quick overview but should not replace reading the full patent text for detailed analysis. Second, as of July 2025, the practices and metrics used to evaluate the accuracy of AI-generated summaries are still an active area of ongoing research. However, there are some promising pipelines for evaluating AI-generated summaries such as the one we have adopted which is named G-Eval. Put simply, G-Eval uses LLM to evaluate the quality of summaries without a ground truth. Using this approach, we have fed 12 patent applications and their corresponding AI-generated summaries to Gemini which judged and rated their quality based on four key dimensions: coherence (4.8), consistency (4.9), fluency (3.0), and relevance (4.8) (scale of 1-5 where higher is better). So while the fluency score is a bit on the low side, which may indicate the awkward phrasing used in the original patent applications, other dimensions scored almost perfectly. We should also note that we did employ some (limited) human evaluation of these summaries and judged them to be adequate for cataloging and general use purposes.

The search option lists applicants according to the distribution, from the highest to lowest. The dropdown list you see contains only the first 20 most active applicants. If you know the applicant’s name, you can search for it using the regular or advanced search options.

For 34 applications (or patents), there are no files. This omission was in the original registry.

• Yes. We added tags of Kind of Product or Services. This is our categorization. You may categorize otherwise, of course.

• We categorized many inventions in more than one category.

• • There were three channels to apply for a Palestine Patent:

(1) Local application. For these there is nothing to indicate that they are in the other options.

(2) An application based on a British patent. If you wish to search for these, select “British Registration” (272 such applications)

(3) An application based on one that was submitted in a country that was a member of an international treaty, in which case the priority date was that of the foreign application. For these, you may select the countries you are interested in. Note that there were 175 such applications that cited ‘Great Britain’ as the priority country. These should not be confused with the previous channel.

• If you notice any errors, please let us know, using this email.  Your contributions will help us enhance the registry’s quality and ensure a more reliable resource for future users. Thanks!