# US birth times, 2025, and the birth chart: data and code

Data and code behind the ZodiScope study at
https://www.zodiscope.io/journal/what-time-are-most-babies-born

Free to reuse with credit to ZodiScope (zodiscope.io).

## 1. CDC counts (real records)

Source: NCHS natality public-use file for 2025, US births (50 states and DC).

| What | File | SHA-256 |
|---|---|---|
| Data file | https://ftp.cdc.gov/pub/Health_Statistics/NCHS/Datasets/DVS/natality/Nat2025US.zip (231,711,170 bytes). Member `Nat2025PublicUS.c20260624.r20260922.txt` | `f5de5435144780919d00d73f508bfb2d5ac1589817a7e6f9173e6a5065368d88` |
| User Guide (record layout, PDF pages 11 and 29-32) | https://ftp.cdc.gov/pub/Health_Statistics/NCHS/Dataset_Documentation/DVS/natality/UserGuide2025.pdf | `6bebb300e2ab0e3fec7293c468e44f8f187233833e4962fa1d824745c79430d8` |

Fields used (1-based byte positions): birth year 9-12, birth month 13-14,
time of birth 19-22 (hhmm, 24-hour local clock time, 9999 = not stated), day of
week 23 (1 = Sunday), delivery method recode DMETH_REC 408 (1 vaginal,
2 cesarean, 9 not stated). `cdc_aggregate.py` also reads birth place (32),
residence status (104) and induction of labor (383) for checks.

All tables use every record in the file: 3,619,990 births occurring in the US
in 2025, including 7,978 births to foreign residents. NCHS reports count US
residents only (3,612,012, which this file matches exactly); the difference
moves no hourly share by more than 0.004 percentage points. Shares by hour and
minute use the 3,619,476 records with a stated time (99.986%).

Rebuild:

```
mkdir -p <dir>/work
unzip -p Nat2025US.zip | python3 -I cdc_aggregate.py Nat2025PublicUS.c20260624.r20260922.txt <dir>/work
python3 -I cdc_build_outputs.py <dir>
```

(The zip uses Deflate64, which Python's zipfile cannot read, so `unzip -p`
streams it.) Outputs: `month_minute.csv` (month x clock minute),
`month_minute_by_method.csv`, `weekday_hour.csv`, `summary.json`.

## 2. Sky model (model figures, not counts of people)

`sky_model.py` takes `month_minute.csv` and casts the sky for every recorded
clock minute with the NASA JPL DE440 ephemeris (`de440s.bsp`, from
https://ssd.jpl.nasa.gov/ftp/eph/planets/bsp/) through Skyfield 1.55, tropical
zodiac.

- The public file has no state and no day of the month. Each month's births
  are spread evenly over the days of that month in 2025, and all births are
  placed in one city at a time (13 cities, city-hall coordinates, each city's
  own time zone and 2025 daylight-saving rules). A city figure means "births
  timed like the US in 2025, born in this city".
- Ascendant and Midheaven from apparent sidereal time and the true obliquity.
- Day chart: the Sun between the Ascendant and the Descendant through the
  Midheaven (houses 7 to 12).
- Sun quadrant: the Ascendant, Midheaven, Descendant and IC bound the same
  four quadrants in every house system whose 1st and 10th cusps are the
  Ascendant and Midheaven (Placidus, Koch, Porphyry, Regiomontanus, Campanus,
  Alcabitius, Topocentric), so these shares do not depend on which of them
  is used.
- Noon chart: the Moon's sign at 12:00 local clock time on the same date.

Check: run `python3 sky_model.py --validate --ephem de440s.bsp`. It reproduces
the sky-time figures published on https://www.zodiscope.io/calculators/birth-chart
(New York, 2026: a rising sign change 15, 30 and 60 minutes later at 12.5%,
25.1% and 50.1% of minutes).

Run:

```
python3 sky_model.py --counts month_minute.csv --by-method month_minute_by_method.csv \
    --year 2025 --ephem de440s.bsp --out sky_model_results.json
```
