This repository contains structured tabular data extracted from the historical Rosenwald medical directories, spanning from 1887 to 1906. The dataset provides valuable insights into the French medical profession during the late 19th and early 20th centuries.
The Rosenwald medical directories were comprehensive annual listings of medical practitioners in France, documenting physicians across Paris and the provinces. This dataset represents a machine-readable extraction of those historical records.
- Total files: 4,166 TSV files
- Time period: 1887-1906 (20 years)
- Total entries: ~381,000 lines (including headers)
- Geographic coverage: Paris and French provinces
- Data format: Tab-separated values (TSV)
Each TSV file represents a single page from the original Rosenwald directories and follows this schema:
| Column | Description |
|---|---|
nom |
Physician's surname/family name |
année |
Year of medical degree or qualification |
notes |
Professional notes, specialties, honors (e.g., ✶ for decorations, "mal. des yeux" for eye diseases, "dent." for dentist) |
adresse |
Practice address (street name and number) |
horaires |
Consultation hours (e.g., "Lun. Mer. Ven. 3 à 5" = Monday, Wednesday, Friday, 3-5pm) |
sexe |
Gender (when specified) |
Files are named using the pattern: YYYY-NNNN.tsv
YYYY: Publication year (1887-1906)NNNN: Page number (zero-padded, 4 digits)
Example: 1887-0022.tsv represents page 22 from the 1887 directory.
nom année notes adresse horaires sexe
Abadie 1870 ✶, Mal. des yeux. Volney 9 Lun. Mer. Ven. 3 à 5
Adam 1875 dent. pl. St-Michel 1
Agard 1870 Bondy 28 1 à 3
Anger (B.) 1865 ✶, Ex-Int. des Hôp. boul. Haussmann 33 3 à 6 The data was extracted using a multi-stage pipeline combining OCR (Tesseract) and AI-powered text analysis (Gemini). The original source material consists of scanned pages from historical Rosenwald medical directories.
The copy.sh script demonstrates the data collection workflow:
# Copy extracted TSV files from the analysis directory
cp /path/to/analysed-tsv/*.tsv rosenwald-extraction/import pandas as pd
import glob
# Load a single file
df = pd.read_csv('rosenwald-extraction/1887-0022.tsv', sep='\t')
# Load all files from a specific year
files_1887 = glob.glob('rosenwald-extraction/1887-*.tsv')
df_1887 = pd.concat([pd.read_csv(f, sep='\t') for f in files_1887])
# Load entire dataset
all_files = glob.glob('rosenwald-extraction/*.tsv')
df_complete = pd.concat([pd.read_csv(f, sep='\t') for f in all_files])# Count physicians by year of qualification
print(df_complete['année'].value_counts().sort_index())
# Find specialists
specialists = df_complete[df_complete['notes'].notna()]
# Extract Paris addresses
paris = df_complete[df_complete['adresse'].notna()]This dataset can be used for:
- Historical demographics: Studying the growth and distribution of the medical profession in France
- Medical specialization: Tracking the emergence of medical specialties over time
- Urban geography: Mapping medical practices in Paris and provincial cities
- Gender studies: Analyzing women's entry into the medical profession
- Social history: Understanding professional honors, hospital affiliations, and consulting hours
- Network analysis: Identifying medical families and professional connections
- Some fields may be empty (especially
sexewhich was rarely documented) - Addresses use historical street names and numbering systems
- Medical specialties are often abbreviated or in French
- Symbols like ✶ (star), ✢, and O indicate various honors and decorations
- OCR and extraction errors may be present despite multi-stage processing
For questions, feedback, or collaboration:
Ren Yi
Email: renyi1006@gmail.com
For more detailed information, see the full report:
- Google Drive folder: https://drive.google.com/drive/folders/1p7XqrWZg4O1lyLGvgXd4mwM9ZVFZBUEy