Some publishers have retractions because one journal had a catastrophe. Others have retractions because they publish across a large, busy scientific continent.
ACS belongs mostly to the second category.
In the Retraction Watch dataset uploaded, ACS-published journals are one of the largest society-publisher clusters. But the interesting point is not merely that ACS has many retractions. The interesting point is why. The answer is not a single scandal, not one rogue journal, not one dominant country, and not a paper-mill flood. It is a mixture of portfolio size, chemistry’s evidence culture, materials-science image/data problems, broad international authorship, and a recent shift toward newer ACS open-access and applied journals.
A caveat first: this analysis uses Retraction Watch records, not total publication denominators. So it can tell us how many retraction records appear in the database, and what their patterns are. It cannot directly tell us the true probability that a paper in ACS, RSC, ASBMB, or AACR will be retracted.
Still, the patterns are revealing. The retraction ledger has fingerprints. 🧪📉
The dataset and classification
I used the uploaded Retraction Watch CSV and focused on records with valid original-publication and retraction-notice dates. To avoid conference-proceedings distortions, I excluded conference abstracts/papers and conference-like proceedings titles. I then classified high-confidence society publishers using the publisher field.
This produced 3,158 non-conference society-publisher records.
ACS had:
| ACS measure | Value |
|---|---|
| ACS retraction records | 418 |
| ACS journals represented | 52 |
| Median time to retraction | 1.68 years |
| Mean time to retraction | 2.89 years |
| Records retracted after 10 years | 5.5% |
| Records involving data/results/method concerns | 71.8% |
| Records involving image concerns | 36.6% |
| Records involving fraud/misconduct tags | 21.5% |
| Records involving paper-mill/peer-review tags | 0.5% |
ACS is a large nonprofit society publisher. ACS Publications says it offers more than 80 hybrid and open-access titles, and its author-facing journal list is correspondingly broad across chemistry and related fields. That matters. A society with 80-plus titles has a much larger “surface area” for retractions than a society with three or ten journals.
Hypothesis 1: ACS has many retractions because it has many journals
Mostly supported.
In the uploaded data, ACS had retraction records spread across 52 different ACS journals. That was the highest number of society-linked journals represented in this filtered society subset.
By comparison:
| Society publisher group | Retraction records | Journals represented in dataset | Records per represented journal |
|---|---|---|---|
| RSC | 539 | 40 | 13.5 |
| ASBMB/JBC | 452 | 4 | 113.0 |
| ACS | 418 | 52 | 8.0 |
| Portland/Biochemical Society | 331 | 3 | 110.3 |
| AACR | 233 | 9 | 25.9 |
| ASM | 211 | 16 | 13.2 |
| AAAS/Science | 187 | 5 | 37.4 |
| IEEE | 160 | 21 | 7.6 |
| BMJ/BMA | 153 | 37 | 4.1 |
| AHA | 150 | 10 | 15.0 |
Non-conference society-linked records. ACS has many retractions, but they are distributed across many journals rather than concentrated in one or two titles.
Calculated from the uploaded Retraction Watch CSV.
This is the first key to ACS. Its retractions are wide, not deep.
ASBMB and Portland Press/Biochemical Society have almost as many or more records than ACS, but those records are concentrated in very few journals, especially biochemical/life-science outlets. ACS, in contrast, has a broad portfolio, so the retraction burden is distributed over many titles.
RSC also has a large chemistry portfolio, officially 57 peer-reviewed journals, and in this dataset it has more society-publisher retraction records than ACS. AACR, by contrast, has 10 peer-reviewed journals, and ASBMB’s core journal program is much smaller, centered around journals such as JBC, Molecular & Cellular Proteomics, and Journal of Lipid Research.
So yes, portfolio size matters. ACS has many retractions partly because ACS publishes many journals.
But that is only the front door. The interesting rooms are inside.
Hypothesis 2: ACS retractions are driven by fraud more than other society publishers
Not supported.
ACS is not unusually fraud-heavy compared with the rest of the society-publisher subset.
Compared with all other society-publisher records:
| Reason theme | ACS | Other society publishers | Odds ratio | p-value |
|---|---|---|---|---|
| Data/results/method concerns | 71.8% | 57.5% | 1.88 | 4.4e-08 |
| Image concerns | 36.6% | 48.0% | 0.63 | 1.7e-05 |
| Fraud/misconduct | 21.5% | 23.8% | 0.88 | 0.34 |
| Plagiarism/duplication/copyright | 28.5% | 36.2% | 0.70 | 0.0026 |
| Paper-mill/peer-review/AI | 0.5% | 10.8% | 0.04 | 2.9e-11 |
This is important. ACS does not look like a paper-mill publisher in this dataset. Only 0.5% of ACS records had paper-mill/peer-review/AI-type tags, compared with 10.8% among other society publishers.
Its strongest enrichment is instead:
Data, results, methods, materials, and reliability concerns.
That fits chemistry. Chemistry papers often depend on spectra, structures, synthetic routes, materials characterization, catalytic activity, nanomaterial images, electrochemical data, crystallographic measurements, and computational/analytical reproducibility. The weak point is often not “fake peer review.” It is whether the data and interpretation hold.
Hypothesis 3: ACS has many retractions because chemistry and materials science are retraction-prone subjects
Partly supported, but better phrased as “ACS sits in evidence-rich, data-fragile subjects.”
ACS retractions are overwhelmingly physical-sciences and engineering-linked.
| Subject flag in ACS records | Share of ACS records |
|---|---|
| Physical sciences / engineering | 77.5% |
| Biology / life sciences | 34.2% |
| Health sciences / medicine | 13.4% |
| Environmental sciences | 4.5% |
| Business / technology / computing | 2.9% |
The broad subject pattern explains much of the ACS profile. ACS is not primarily a clinical-journal publisher, not a social-science publisher, and not a conference-proceedings publisher. It lives in the chemical sciences, materials science, analytical chemistry, nanoscience, chemical biology, medicinal chemistry, catalysis, polymers, and environmental chemistry.
ACS Publications itself describes its journals as spanning the chemical sciences and related fields, and its submission guidance emphasizes choosing the appropriate journal across a broad spectrum of scientific disciplines.
The retraction reason profile follows the subject profile:
| ACS subject cluster | Records | Median lag | Image concerns | Plagiarism/duplication | Paper-mill/peer-review |
|---|---|---|---|---|---|
| Physical sciences / engineering | 324 | 1.62 years | 37.7% | 28.7% | 0.0% |
| Biology / life sciences | 143 | 2.01 years | 42.0% | 29.4% | 1.4% |
| Health sciences / medicine | 56 | 1.83 years | 51.8% | 37.5% | 0.0% |
So ACS retractions are not dominated by old clinical fraud or modern special-issue paper mills. They are dominated by data-heavy chemistry and materials-science problems.
Hypothesis 4: ACS retractions are concentrated in only a few bad journals
Not really. There are important clusters, but the distribution is broad.
The top ACS journals by retraction records were:
| ACS journal | Records | Median lag | Data concerns | Image concerns | Fraud/misconduct | Plagiarism/duplication |
|---|---|---|---|---|---|---|
| ACS Omega | 55 | 1.84 years | 74.5% | 70.9% | 30.9% | 50.9% |
| Journal of the American Chemical Society | 54 | 2.81 years | 85.2% | 16.7% | 27.8% | 7.4% |
| ACS Applied Materials & Interfaces | 32 | 2.40 years | 75.0% | 43.8% | 34.4% | 34.4% |
| Organic Letters | 29 | 1.75 years | 79.3% | 24.1% | 10.3% | 6.9% |
| The Journal of Organic Chemistry | 17 | 0.33 years | 58.8% | 5.9% | 0.0% | 29.4% |
| Industrial & Engineering Chemistry Research | 17 | 2.63 years | 64.7% | 58.8% | 29.4% | 52.9% |
| Biochemistry | 16 | 1.56 years | 87.5% | 6.3% | 31.3% | 6.3% |
| Langmuir | 16 | 1.51 years | 68.8% | 43.8% | 25.0% | 37.5% |
| Journal of Chemical & Engineering Data | 15 | 0.93 years | 86.7% | 13.3% | 6.7% | 13.3% |
| ACS Nano | 13 | 2.60 years | 61.5% | 61.5% | 7.7% | 53.8% |
Top ACS journals by Retraction Watch records. ACS Omega and JACS have similar counts but different reason profiles.
Calculated from the uploaded Retraction Watch CSV.
The top two titles tell two different stories.
ACS Omega is a broad open-access journal with a recent retraction cluster. Its records are heavily image- and duplication-linked.
JACS is ACS’s flagship chemistry journal. Its retractions are more data/results-heavy and less image-duplication-heavy. The median lag is longer, suggesting a more traditional high-visibility correction pathway.
That distinction matters. “ACS retractions” are not one species. They are a small zoo.
Hypothesis 5: ACS retractions have increased recently because ACS expanded its portfolio and newer titles contribute more
Supported by the data, with caution.
ACS records increased sharply in the most recent era.
| Era | ACS records | ACS journals represented | Median lag | Image concerns | Data concerns |
|---|---|---|---|---|---|
| ≤2009 | 36 | 11 | 1.15 years | 5.6% | 66.7% |
| 2010–2014 | 63 | 22 | 1.31 years | 27.0% | 68.3% |
| 2015–2019 | 129 | 34 | 1.31 years | 26.4% | 65.1% |
| 2020–2026 | 190 | 41 | 2.19 years | 52.6% | 78.4% |
ACS Omega appears mainly in the 2020-2026 era, changing the ACS retraction profile.
Calculated from the uploaded Retraction Watch CSV. The 2020-2026 era includes partial 2026.
The major shift is ACS Omega. In this dataset, 53 of the 55 ACS Omega retractions occurred in the 2020–2026 notice era. This does not automatically mean ACS Omega is less reliable. It may reflect its youth, size, breadth, open-access scale, recent integrity screening, or a combination of these.
The overall ACS shift after 2020 also shows rising image concerns: 52.6% of ACS records in 2020–2026 had image-concern tags, compared with 26–27% in 2010–2019.
That is one of the clearest internal ACS trends.
Hypothesis 6: ACS retractions are country-driven
Partly supported, but not in a simple way.
ACS records are strongly concentrated among China, the United States, and India, but the reason profiles differ.
| Country in ACS records | Records | Median lag | Image concerns | Fraud/misconduct | Plagiarism/duplication |
|---|---|---|---|---|---|
| China | 140 | 1.08 years | 30.7% | 7.9% | 33.6% |
| United States | 132 | 2.57 years | 24.2% | 25.8% | 12.9% |
| India | 55 | 1.98 years | 60.0% | 29.1% | 50.9% |
| France | 21 | 3.77 years | 47.6% | 9.5% | 28.6% |
| Japan | 20 | 2.14 years | 30.0% | 40.0% | 5.0% |
| Saudi Arabia | 20 | 2.18 years | 75.0% | 55.0% | 75.0% |
Country-paper occurrences in ACS records. Multi-country papers count once for each listed country.
Calculated from the uploaded Retraction Watch CSV.
China and the United States have similar record counts in ACS, but different correction clocks.
China-linked ACS records are corrected faster, median 1.08 years, and have lower fraud/misconduct tagging.
United States-linked ACS records are slower, median 2.57 years, and have higher fraud/misconduct tagging.
India-linked ACS records are fewer but more image- and plagiarism/duplication-heavy.
This suggests that ACS is not being shaped by one country pipeline. It sits at the intersection of several chemistry communities with different correction signatures.
Hypothesis 7: ACS retractions are mostly single-country, not multinational
Supported.
In ACS records:
| Collaboration type | Records | Median lag | Image concerns |
|---|---|---|---|
| Single-country | 311 | 1.53 years | 35.7% |
| Multinational | 107 | 2.07 years | 39.3% |
About 25.6% of ACS records were multinational. The median lag was slightly longer for multinational records, but the Mann-Whitney test was not significant in this dataset (p = 0.20). So multinational status is not a strong ACS-specific driver.
The most common exact country sets were simple:
| Exact country set | ACS records |
|---|---|
| China | 98 |
| United States | 90 |
| India | 41 |
| Japan | 16 |
| China + United States | 15 |
| France | 10 |
This again points away from a single international-collaboration explanation. ACS retractions are broad, not dominated by one exotic collaboration pattern.
Hypothesis 8: ACS differs from RSC and other chemistry society publishers
Supported.
ACS and RSC are the obvious comparison because both are large chemistry society publishers. RSC officially publishes 57 peer-reviewed journals, while ACS reports more than 80 titles.
In the uploaded Retraction Watch data:
| Publisher | Records | Journals represented | Median lag | Image concerns | Data concerns | Paper-mill/peer-review |
|---|---|---|---|---|---|---|
| RSC | 539 | 40 | 2.40 years | 59.7% | 81.1% | 15.0% |
| ACS | 418 | 52 | 1.68 years | 36.6% | 71.8% | 0.5% |
A Mann-Whitney test shows ACS retractions are significantly faster than RSC retractions in this filtered society subset (p = 1.07e-05). The effect size is modest, but real.
ACS also differs from the slow biomedical society publishers:
| Comparison | Other society median | ACS vs other result |
|---|---|---|
| ASBMB/JBC | 7.09 years | ACS much faster, p = 1.8e-57 |
| AACR | 8.67 years | ACS much faster, p = 2.9e-41 |
| ASM | 4.28 years | ACS faster, p = 2.1e-12 |
| AHA | 6.34 years | ACS faster, p = 3.3e-17 |
| AAAS/Science | 1.97 years | No strong difference, p = 0.18 |
| AIP | 1.49 years | No strong difference, p = 0.57 |
| APS | 0.94 years | ACS slower, p = 0.00088 |
This places ACS in the middle:
- Faster than biomedical image-forensics societies.
- Slower than some physics/computing process-correction clusters.
- Similar to broad high-visibility or physical-science society publishers.
- Less paper-mill/peer-review-heavy than RSC in this dataset.
The main reason ACS has many retractions: breadth plus data-rich chemistry
Let us now answer the central question directly.
ACS journals have many retractions in the uploaded data because of six overlapping factors.
1. ACS has a very large journal portfolio
ACS has more than 80 titles, which is larger than many society publishers. More journals mean more total publication surface area, and therefore more possible retraction records.
But unlike ASBMB or AACR, ACS’s retractions are not concentrated in a tiny number of journals. They are spread across 52 ACS journals in the Retraction Watch data.
2. ACS publishes in data-dense fields
Chemistry and materials science generate many retraction-relevant evidence types:
- spectra,
- chromatograms,
- crystallographic data,
- compound characterization,
- electrochemical curves,
- catalytic activity measurements,
- microscopy images,
- nanomaterial figures,
- computational chemistry outputs,
- biological assay panels in chemical biology and medicinal chemistry.
This produces many opportunities for data/results/method concerns. In the data, 71.8% of ACS records had data/results/method-related tags.
3. ACS has a strong materials and nanoscience footprint
The ACS journals with substantial retraction counts include ACS Applied Materials & Interfaces, ACS Nano, Langmuir, Industrial & Engineering Chemistry Research, ACS Sustainable Chemistry & Engineering, and ACS Omega. These fields often depend on images and materials characterization.
That likely explains why ACS image concerns rose sharply in the 2020–2026 era.
4. ACS Omega changed the recent profile
ACS Omega alone accounts for 55 records, mostly in 2020–2026. It has a very different profile from JACS: more image concerns, more duplication/plagiarism tags, and a recent surge.
ACS Omega therefore acts as a recent amplifier in the ACS retraction profile.
5. ACS is international, but not dominated by one country
China and the United States each contribute large ACS country-paper counts, while India is third. But their signatures differ. China-linked ACS records are faster and more duplication-associated; US-linked records are slower and more fraud/misconduct-associated; India-linked records are more image-heavy.
The country pattern is real, but no single country explains ACS.
6. ACS does not show a strong paper-mill/peer-review signature
This is crucial. Unlike some other large retraction clusters, ACS records are not primarily marked by compromised peer review or paper-mill themes. The ACS story is not mainly “fake review factories.” It is mostly data reliability, images, duplication, and chemical evidence integrity.
Why ACS is not like ASBMB, AACR, or RSC
ACS versus ASBMB/JBC
ASBMB/JBC records are concentrated in a few journals and are slow, with a median lag of 7.09 years. These records are nearly all biology/life-science and heavily image/data/misconduct-linked.
ACS is broader and faster. ACS is not a long-tail western-blot archive in the same way.
ACS versus AACR
AACR records are cancer-biology-heavy, with a median lag of 8.67 years and high image/fraud/misconduct shares. AACR’s 10 journals cover cancer research from basic biology to clinical and epidemiological research.
ACS has fewer very-late biomedical forensic corrections. Its median is much shorter.
ACS versus RSC
RSC is ACS’s closest chemistry cousin. RSC has more records in this dataset, higher image-concern share, and much higher paper-mill/peer-review tagging than ACS. RSC’s median lag is also longer.
ACS therefore looks less like a batch-correction/paper-mill case than RSC does in this dataset.
What ACS should worry about, based on the data
The data do not say ACS has a fraud crisis. They say ACS has a data integrity and figure integrity burden.
The highest-priority risk zones are:
- Data/results/method reliability
- ACS’s strongest enriched reason theme.
- Especially important in JACS, Organic Letters, Biochemistry, Journal of Chemical & Engineering Data, and materials journals.
- Image integrity in materials/nano/open-access titles
- Strong in ACS Omega, ACS Nano, ACS Applied Materials & Interfaces, Industrial & Engineering Chemistry Research, Langmuir, ACS Applied Bio Materials.
- Duplication and overlap
- Prominent in ACS Omega, ACS Nano, Industrial & Engineering Chemistry Research, ACS Sustainable Chemistry & Engineering.
- Recent growth and broad-scope journals
- ACS Omega changed the 2020–2026 profile substantially.
- International heterogeneity
- Country signatures differ enough that editorial screening may need field- and region-aware support, without turning that into crude national profiling.
ACS’s own peer-review guidance notes that submissions undergo screening and editor assessment for scope, scientific quality, and impact before formal peer review. The Retraction Watch patterns suggest that the next layer of defense is not just ordinary peer review. It is technical screening: image forensics, data availability, compound characterization checks, spectral validation, crystallographic validation, and stronger policies around raw data.
Final diagnosis: ACS has many retractions because it is large, broad, chemical, and data-rich
The tempting headline would be:
“ACS journals have many retractions.”
The better headline is:
“ACS journals have many retractions because ACS publishes a large, broad, data-intensive chemistry portfolio, and its retraction problems are distributed across many journals rather than concentrated in one catastrophic title.”
ACS is not the worst society publisher in raw count here. RSC and ASBMB/JBC have more records in this filtered society subset. ACS is also not unusually paper-mill-heavy. Its retractions are mostly not the industrial fraud pattern seen in some modern journal clusters.
ACS’s problem is more classical and more chemically specific:
- data that do not hold,
- images that raise concerns,
- duplicated or overlapping material,
- materials and nano-characterization problems,
- broad open-access title growth,
- international publication volume,
- and the sheer size of the ACS journal estate.
In other words, ACS does not look like a burning house. It looks like a large chemical city with many laboratories, many instruments, many figures, many spectra, and many chances for the literature’s smoke alarms to go off.
The lesson is not that ACS is uniquely unreliable. The lesson is that large, prestigious, society-run portfolios still need industrial-strength integrity infrastructure.
A society journal is not a magic shield. It is a public trust machine. ACS has built one of the largest such machines in chemistry. The retractions show where the gears grind. 🔬📊
