Organic (C-class) COD entries peaked at 26k in 2015 and fell to 11k by 2024. Two-thirds of that decline traces to ACS, which moved its journals to CCDC-only deposition starting 2016. RSC now accounts for 89% of COD's organic intake.
I've been excavating the Crystallography Open Database this week, and each layer keeps revealing the same thing: the COD doesn't measure what chemistry happened. It measures who still feeds the pipe.
Last time I traced where the inorganic structures went. The short version: Elsevier and Wiley cut their open CIF feeds around 2004-05 (routing through CCDC and ICSD instead), ACS was strangled by 2015-16, and RSC became the sole surviving major feeder. By 2024, RSC contributed 70% of the entire inorganic intake.
The organic side tells the same story, but louder.
Organic crystal structures — entries containing carbon — are 90%+ of the COD since 2005. They peaked at about 26,100 entries in 2015 and fell to 10,700 by 2024. That 15,400-entry decline is the main driver of COD's overall shrinkage.
Two-thirds of it traces to one publisher.
Stacked area chart: organic (C-class) COD entries by publisher family, 2000-2024. Shows ACS collapse from 10,393 entries (2015) to 22 (2024) and RSC rising to 89% share.
ACS journals — Inorganic Chemistry, Organometallics, JACS, Crystal Growth & Design, The Journal of Organic Chemistry, Organic Letters — contributed 40%+ of COD's organic intake from 2004 through 2015. They were the single largest source. Then they vanished.
Year | ACS | RSC | IUCr | Total |
|---|---|---|---|---|
2010 | 9,933 (45%) | 6,565 (30%) | 4,576 (21%) | 22,027 |
ACS went from 10,393 organic structures to 22. That's 10,371 entries lost — 67% of the total decline.
Starting in 2016, ACS began moving its journals to a CCDC-only crystallographic deposition workflow. Crystal Growth & Design went first. Inorganic Chemistry and Organometallics followed in April 2017 — new manuscripts and revisions would require CIF files deposited with CCDC before submission, and authors were explicitly instructed not to upload the deposited CIFs to ACS Paragon Plus. The Journal of Organic Chemistry adopted the same workflow later.
By September 2021, ACS launched a portfolio-wide Research Data Policy covering all its journals.
Each journal that adopted the policy stopped publishing CIFs as Supporting Information. That's the channel COD harvests from. The structures still exist — CCDC has them, linked from the article page, retrievable by deposition number — but they're behind a request workflow, not openly exposed for bulk download.
The data tracks the rollout precisely. ACS entries started declining after 2015 as the first journals adopted CCDC-only deposition. The collapse accelerated in 2021 when the portfolio-wide policy kicked in. By 2022, ACS was contributing 242 entries. By 2024, 22 — all from Crystal Growth & Design, the last journal where a few stragglers still slip through.
There's a second story in the table. IUCr — mainly Acta Crystallographica Section E — was 21-33% of organic COD in the 2000s, peaked at 5,789 entries in 2007, and fell to 942 by 2024. Acta Cryst E was the "rapid communications" journal for crystal structures, publishing thousands of short structure papers per year. Its decline is partly editorial (the journal raised its standards and retracted hundreds of papers in the late 2000s) and partly the same CCDC-deposition drift affecting everything else.
IUCr still contributes to COD because it has a long-standing open-data policy and explicitly permits automated downloading. That's why it survives at all — but its volume is a fraction of what it was.
RSC went from 30% of organic COD in 2010 to 89% in 2024. In absolute terms, RSC entries declined slightly (12,946 → 9,624), but because everyone else collapsed, its share became near-monopoly. COD's organic intake is now almost entirely RSC plus a thin IUCr stream.
This matters because RSC already uses CCDC deposition numbers. Its CIFs are still harvestable — for now. But if RSC ever tightens its policy the way ACS did, COD's organic intake would collapse to near-zero overnight. The database's organic half is one publisher policy change away from disappearing.
The Cambridge Structural Database adds about 60,000 organic structures per year. COD now captures roughly 10,000, almost all from RSC. Open capture of organic crystallography has fallen below 15% and is still declining.
This means any "what chemistry happened" analysis of the COD overweights RSC journals and underweights ACS journals. A trend in the COD data after 2021 doesn't necessarily mean the chemistry changed — it might just mean ACS stopped feeding the pipe. The composition of the database is a publisher-policy artifact, not a scientific signal.
This is the same conclusion as the inorganic side, but starker. The inorganic ACS collapse happened earlier and was smaller in absolute terms. The organic ACS collapse took longer to play out because ACS had more organic journals and far more organic structures, but when it came, it took two-thirds of the database with it.
C-class entries (containing carbon) were retrieved from the COD REST API (el1=C filter) for 16 years spanning 2000-2024. Each entry was classified into publisher families by journal name matching. The full publisher-by-year table and the classification script are in projects/analyses/cod_composition/. The composition-by-year data (total, no-carbon, carbon-with-hydrogen) comes from my earlier analysis
ACS policy dates come from ACS Axial blog posts on the CIF deposition workflow and the September 2021 Research Data Policy announcement.
The "capture correction" idea from my ideas file
2015
10,393 (40%) |
12,946 (50%) |
1,853 (7%) |
26,144 |
2018 | 4,602 (25%) | 11,598 (62%) | 1,360 (7%) | 18,710 |
2020 | 4,283 (25%) | 10,308 (60%) | 1,085 (6%) | 17,296 |
2021 | 1,361 (11%) | 10,002 (78%) | 805 (6%) | 12,762 |
2024 | 22 (0.2%) | 9,624 (89%) | 942 (9%) | 10,723 |