Journal of Interesting Results with Terrible Sample Sizes

┤ §1 · Manifesto ├

Most findings are probably nothing. Publish them anyway.

“i would support an academic journal called Probably Nothing where you can publish interesting results with terrible sample sizes”
The founding line. We built the whole journal around it.

Science has a file-drawer problem: the intriguing pilot, the n=1 that made the whole lab go hmm, the correlation too funny to be meaningless and too fragile to be anything else — all of it dies in a drawer because no respectable journal will touch a result with twelve subjects and a p-value of 0.11. We are not a respectable journal.

Probably Nothing exists for results that are interesting but underpowered: too small to trust, too delightful to suppress. Every paper states its n up front, in the title-adjacent metadata, where it cannot hide. Every p-value is reported with shame — exact, unrounded, and accompanied by the authors' own assessment of how little it means.

Our editorial position, stated formally: n=1 is celebrated here. A single careful observation, honestly reported, beats a thousand unexamined ones. It also beats most of our back catalog, which currently stands at seven papers, five of them at n=1, concerning other people's patients, one baby, and seven stars.

┤ §2 · Aims & Scope ├

What we publish, and what even we won't touch.

The journal's scope is interesting results with terrible sample sizes. “Interesting” is judged by whether at least one editor says “huh!” out loud. “Terrible” is judged statistically. Both bars are enforced with equal unseriousness.

Submission	Verdict	Why
Fascinating effect, n=1 (yourself)	IN SCOPE · CELEBRATED	Autoethnography is just n=1 with confidence.
Promising pilot, n=7, p=0.09	IN SCOPE	Report the p-value with shame (see §4, rule 3).
Null result, n=400, impeccable methods	DESK REJECT	Your sample size is showing. Try a journal with standards.
Boring result, n=3	DESK REJECT	Terrible alone is not enough. It must also be interesting.
p=0.049, n=2,000, undisclosed peeking	SHAME REJECT	Adequately powered and still suspicious. Impressive. No.

EDITORIAL POLICY ON p-VALUES All p-values must be reported exact to two decimals, in a footnote beginning “we are aware”. The sentence “marginally significant” is banned. The sentence “trending toward significance” is banned twice.

┤ §3 · Current Issue ├

Vol. 0, No. 0 — Inaugural Issue

7 articles · combined n=19 · units inconsistent · 0 replications attempted, 0 encouraged

PN-2510.00001 [q-bio.NC] · v1 07 Oct 2026 · n=1

Loss of Recent Memory After Bilateral Hippocampal Lesions

W. B. Scoville, B. Milner — Dept. of Neurosurgery, Hartford Hospital; Montreal Neurological Institute

Abstract — In 1953, a 27-year-old man with intractable epilepsy (H.M.) underwent bilateral medial temporal-lobe resection. He awoke with his intellect and childhood memories intact and no ability to form new ones — a deficit the authors document across ten operated cases, of which this one was total. The paper reports no p-values, having none to be ashamed of, and founds modern memory neuroscience on a single hippocampus-shaped hole. Limitations include n=1, and that the lesion was, in retrospect, somewhat difficult to undo.

Sample size: n=1 index case (H.M.; 9 comparator resections surveyed in the same paper)

Comments: 11 pages, 2 figures. Data available upon awkward request (the patient kept nothing; the field kept everything).

Subjects: amnesia (n=1) hippocampus case-series core

Citation: Scoville, W. B. & Milner, B. (1957). J. Neurol. Neurosurg. Psychiatry 20, 11–21. doi:10.1136/jnnp.20.1.11 · free full text
PN-2510.00002 [q-bio.NC] · v1 07 Oct 2026 · n=1

Conditioned Emotional Reactions

J. B. Watson, R. Rayner — Psychological Laboratory, Johns Hopkins University

Abstract — Albert B., 9 months old, liked rats. The authors struck a steel bar behind his head whenever he reached for one, until he did not. The fear generalized to rabbits, dogs, fur coats, and Santa Claus masks; the infant was discharged before any de-conditioning, a limitation the authors note and the century has not forgiven. The entire experimental literature of fear conditioning descends from one baby who never consented, in the modern sense or any other.

Sample size: n=1 infant (Albert B., 9 months at intake)

Comments: 14 pages. Ethics approval: none sought, none available; it was 1920.

Subjects: fear conditioning (n=1) behaviorism steel-bar-omics

Citation: Watson, J. B. & Rayner, R. (1920). J. Exp. Psychol. 3(1), 1–14. · free full text
PN-2510.00003 [q-bio.NC] · v1 07 Oct 2026 · n=1

Memory: A Contribution to Experimental Psychology

H. Ebbinghaus — University of Berlin

Abstract — To study memory uncontaminated by meaning, the author memorized thousands of nonsense syllables and tested himself, repeatedly, for years. Sole subject, sole experimenter, sole beneficiary: n=1, and the 1 was him. Results: the forgetting curve (most of it gone within a day), the savings method, and the spacing effect — the entire quantitative study of human memory, founded on one man's evenings with consonants. Pre-registration: none. Replication: everyone, forever.

Sample size: n=1 (the author himself)

Comments: 155 pages in translation. Reviewer 2 was also him, and was lenient.

Subjects: forgetting (n=1) autoethnography with confidence syllables

Citation: Ebbinghaus, H. (1885). Über das Gedächtnis. Leipzig: Duncker & Humblot. (Trans. Ruger & Bussenius, 1913.) · free full text
PN-2510.00004 [q-bio.PE, q-bio.QM] · v1 07 Oct 2026 · n=1

An Inquiry into the Causes and Effects of the Variolae Vaccinae

E. Jenner — Berkeley, Gloucestershire (country surgeon; no affiliation, no ethics committee, considerable nerve)

Abstract — Noting that milkmaids with cowpox escaped smallpox, the author inoculated 8-year-old James Phipps with cowpox matter from a dairymaid's hand (14 May 1796) and then, seven weeks later, with smallpox itself (1 July 1796). The boy did not sicken. The experiment that founded immunology had n=1, no control arm, and a subject who could not consent because it was 1796. The author published the book anyway (1798), and was right.

Sample size: n=1 boy (James Phipps, 8; further cases accrue around him in the book)

Comments: 75 pages, 4 color plates. Impact factor: all of immunology.

Subjects: vaccination (n=1) milkmaid-omics challenge trial (unpermitted)

Citation: Jenner, E. (1798). An Inquiry into the Causes and Effects of the Variolae Vaccinae. London: Sampson Low. · free full text
PN-2510.00005 [q-bio.CB, q-bio.QM] · v1 07 Oct 2026 · n=1

Attempt to Fulfil Koch's Postulates for Pyloric Campylobacter

B. J. Marshall, J. A. Armstrong, D. B. McGechie, R. J. Glancy — Perth, Western Australia

Abstract — Unable to interest an animal model (or a volunteer) in drinking a culture of the curved bacteria colonizing ulcer patients' stomachs, the first author drank it himself. Acute gastritis followed; repeat endoscopy recovered the organism; antibiotics cleared it. Koch's postulates, fulfilled at n=1, with the investigator serving as subject, vector, and endpoints committee. Nobel Prize, 2005 (shared with J. R. Warren). Side effects included vindication.

Sample size: n=1 (the first author; ethics approval: himself)

Comments: 4 pages. Control group declined to participate (there was only him).

Subjects: self-ingestion (n=1) ulcers (bacterial, allegedly) postulates fulfilled

Citation: Marshall, B. J. et al. (1985). Med. J. Aust. 142(8), 436–439. doi:10.5694/j.1326-5377.1985.tb113443.x · PubMed
PN-2510.00006 [astro-ph.IM, physics.hist-ph] · v1 07 Oct 2026 · n=7 ★

A Determination of the Deflection of Light by the Sun's Gravitational Field, from Observations Made at the Total Eclipse of May 29, 1919

F. W. Dyson, A. S. Eddington, C. Davidson — Royal Observatory, Greenwich; Cambridge Observatory

Abstract — Two expeditions, one eclipse, one recently ended war. The trusted Sobral plates captured 7 stars, yielding 1.98″ ± 0.12″ against Einstein's 1.75″ and Newton's 0.87″; the Príncipe plates, shot through storm cloud, corroborated. General relativity was confirmed on barely a dozen stellar images and announced 6 November 1919 to packed societies and, shortly after, the entire planet. Largest effect-to-sample-size ratio in the history of physics.

Sample size: n=7 stars (Sobral 4-inch; plus storm-degraded Príncipe plates)

Comments: 43 pages, 16 eclipse photographs (Príncipe: mostly cloud).

Subjects: starlight (n=7) relativity (confirmed, pending clouds) eclipse logistics

Citation: Dyson, F. W., Eddington, A. S. & Davidson, C. (1920). Phil. Trans. R. Soc. A 220, 291–333. doi:10.1098/rsta.1920.0009 · publisher
PN-2510.00007 [q-bio.NC, stat.AP] · v1 07 Oct 2026 · n=7

Antidepressant Effects of Ketamine in Depressed Patients

R. M. Berman, A. Cappiello, A. Anand, D. A. Oren, G. R. Heninger, D. S. Charney, J. H. Krystal — Yale / Connecticut Mental Health Center

Abstract — Seven depressed patients, one randomized double-blind crossover, one subanesthetic ketamine infusion: mood lifted within 72 hours where saline did nothing, and stayed lifted for days beyond the drip. The entire rapid-antidepressant revolution — esketamine, the clinic down your street — descends from n=7. Author count: also seven, a ratio of one author per subject that this journal considers the gold standard.

Sample size: n=7 (crossover; each subject their own control, as is tradition at this sample size)

Comments: 4 pages. Power analysis not discussed, for reasons that are obvious.

Subjects: ketamine (n=7) crossover (everyone gets everything) 1 author per subject

Citation: Berman, R. M. et al. (2000). Biol. Psychiatry 47(4), 351–354. doi:10.1016/S0006-3223(99)00230-9 · publisher

┤ §4 · For Authors ├

Submission guidelines. Read them; we barely enforce them.

Submissions to editor@probablynothing.minntelligence.fyi as a single PDF, a napkin photograph, or a forwarded thread. By submitting you affirm the result is interesting and the sample size is terrible.

Abstract ≤ 150 words. If the finding cannot survive 150 words, it cannot survive review. Most of ours barely survived the title.
Methods must fit on a napkin. Photograph of said napkin accepted as Figure 1. “We asked around” is a valid sampling frame if honestly labeled.
p-values reported with shame. Exact values, two decimals, in a footnote beginning we are aware. “Marginally significant” is a desk reject; “trending” is two desk rejects.
n=1 celebrated, n≥300 referred elsewhere. Adequately powered studies will be forwarded — without your consent — to journals with standards and waiting times.
Confess your forking paths. Every analysis you tried lives in a footnote titled “things we also tried.” Reviewers check. Reviewer 2 especially.
Data available upon awkward request. You must share data with anyone who asks, but you may make it briefly weird first. A spreadsheet named FINAL_v2_REAL_final.csv satisfies this rule.
Article processing charge: one (1) compliment to the editor. Sincere compliments are fast-tracked. Flattery about our impact factor is rejected as empirically false.

Decisions: accept (rare, celebrated with a terminal bell), revise-and-shrink (cut the n, keep the joke), or reject-with-affection (framed letter, suitable for a lab door). Median turnaround: 6 days, because our reviewers have nothing else to do — see §5.

┤ §5 · Editorial Board ├

Masthead of the willing.

Editor-in-Chief	VACANT. Applications: have eyes, own at least one opinion about p-values. The chair is literally empty; the previous editor left to pursue adequately powered research and is missed.
Managing Editor	VACANT. Duties include answering mail and pretending Vol. 0, No. 1 is “in production.” It is not.
Statistical Reviewer	VACANT. Role currently filled on alternating weeks by a magic 8-ball (surprisingly stringent) and the managing editor's dog (surprisingly lenient).
Reviewer Pool	Two (2) referees, one of whom keeps striking a steel bar behind the authors' heads (see PN-2510.00002). Median review: “lol. accept with minor revisions.” Minor revisions always requested; never once specified.
Byline	Set in monospace, as nature intended. Authors keep all rights, including the right to be gently mocked at the annual banquet (attendance: n=3).

$ subscribe --journal probably-nothing → no newsletter exists. Check back when someone willingly becomes n=1 again. Tip the editors →