How Ranfft database got integrated into The Watchmakers App

Deniss Fedotovs, 09.09.2026., 13:07

As many of you know, Dr. Roland Ranfft spent years building an incredible archive of watch movements on ranfft.de. Soon after his death, his website went down, and several projects emerged using his data - some fairly popular like ranfft.org or EmmyWatch, and at least three or four others less known. While ranfft.org claims "The data and tools from his original site were completely deleted, with no backup remaining.", that isn’t true.

The original Ranfft archive contains around 10,000 caliber entries and 9,000 images, mostly sourced from legacy Flume catalogues. Because many images were reused for similar movements within the same family, the dataset relies heavily on free-text entries divided into sections. Alongside technical specs, Dr. Ranfft included invaluable bench notes - like warnings about left-handed screws - and notes on historical context.

The easiest way to re-host this data is to run an automated parser and publish whatever comes out without checking it - which is exactly what other projects did. The Watchmakers App took a completely different approach.

Our goal was to normalize and structure the dataset to vastly improve its quality without losing a single detail or historical remark made by Dr. Ranfft. This required months of painstaking, full-time manual work alongside technical scripts to ensure no horological context was misinterpreted or lost in translation.

Where automated scraping fails most glaringly is in context. A primary example is ranfft.org, which misattributed movement production years across their entire database by reading the date of a sample watch description as the movement's manufacturing run. If Dr. Ranfft noted a sample watch from "ca. 1945," ranfft.org listed the caliber's production year as 1945 - even if the actual production period was 1922 to 1940.

By cleaning and properly structuring the data from day one, The Watchmakers App established the most reliable caliber filtering system available:

Zero False Positives: Filter for a "moon phase" complication and you get every caliber with that complication (even optional ones) - nothing more, nothing less. Try that on ranfft.org and you get dozens of movements that don't have a moon phase at all.
Granular Attributes: You can stack filters by movement size, manufacturer, country, balance position, escapement type, hands configuration, or movement shape and get exact matches every time.
Interlinked Families: Every entry is fully cross-referenced, allowing you to jump instantly between generations, family variants, or calibers mentioned inside infosheet remarks.

Because ranfft.org is just an uncurated automated scrape of the legacy site, there is literally no caliber or data on ranfft.org from the original Ranfft database that isn't already in The Watchmakers App - except inside our database, those entries are corrected.

However, once we integrated the original archive, the real work began. We realized Dr. Ranfft's database still had massive gaps - especially regarding pre-1940/1950 movements - and contained errors driven by the limited sources available at the time. For example, if a technical sheet noted that numbers 410 to 419 were reserved for a movement family, the original Ranfft database copied all 10 calibers into the system, even though official manufacturer documentation now proves some of those numbers were never actually produced.

So, we decided to systematically refactor the entire legacy database using primary sources. We sourced physical movement compilation books, technical sheets, and parts catalogues - including every single Ebauches SA catalogue published since 1928, along with loose technical documentation and official internal Ebauches SA production lists through 1960.

By comparing these primary documents against physical movements and online horological research, we have been able to:

Identify & Fix Misattributions: Correct long-standing errors for misidentified calibers across multiple brands.
Decouple Grouped Calibers: Bust the common myth that one caliber designation covers multiple bridge configurations. Official factory documentation proves every bridge combination had its own official designation (and usually base plate, including keyless works, shared between family), even if it wasn't always communicated publicly.
Clean Out Duplicate Noise: Eliminate ghost entries for theoretical calibers while keeping full, accurate family listings intact.
Establish Verified Production Dates: Determine accurate production windows based on catalog publication dates and official deprecation lists.
Expand the Image Catalog: Identify reused generic photos and replace them with precise, high-resolution images of the exact caliber.

Because we are aggressively filling historical gaps by adding thousands of early industrial calibers and upgrading reused pictures to exact-caliber photos, our image database has expanded by over 50%, drastically reducing duplicate image confusion when matching unknown movements.

This refactoring isn't a one-off update - it's a full-time daily effort reviewing tens of calibers every day, which is why The Watchmakers App gets database updates several times a month and has become the fastest-growing horological database available.

In the spirit of Dr. Ranfft’s original work, the raw database records, specs, and caliber images remain 100% free for everyone. To support this full-time archival work, our paid subscription plan unlocks the app's advanced technical feature set - including full multi-variable search filtering, photo-based movement identification, technical sheets, and parts interchangeability lists.

back

Join Our Discord