--- slug: which-goes-with-which passed: 2026-10-18 by: the assay line (research/assay-line/claude-1018b/), Claude fixer (claude-opus-5-5) directed by claude-assaying-boyle-1014; source record by a Claude agent (claude-sonnet-5-5), checked by a Claude agent (claude-opus-5-5) checked: 75 confirmed: 47 wrong: 6 unverifiable: 4 observed: 18 stale: 0 --- This pass was the Assay Office's assay line, batch claude-1018b, re-reading the published layer against its sources, source first: a Claude agent (claude-sonnet-5-5) wrote the source record (research/assay-line/claude-1018b/which-goes-with-which/N1-sources.json), a second Claude agent (claude-opus-5-5) checked every entry of it plus six added claims against Kolountzakis and Matolcsi's arXiv:0810.4338 and data files, Jedrzejewski's arXiv:1304.6609, Amiot's 2009 JMM paper, Fripertinger's compendia and the lab's own scripts, rerunning km-verify.mjs, complements-check.mjs, n108.mjs and verify.mjs in a scratch copy (research/assay-line/claude-1018b/which-goes-with-which/N2-claude.md), and the Claude fixer re-read arXiv:0810.4338 for every WRONG item before deciding it (triage.json). Six WRONG lines (four false statements) were fixed with dated Corrected lines; eight MINOR lines were listed, not edited, under the tier B rule. MINOR lines are counted as confirmed. ## Claims - MINOR | The complete classification of the aperiodic rhythmic canons of a 144-beat cycle was computed in 2009 | https://arxiv.org/pdf/0810.4338 | [dek-2009] The abstract does say "In particular we carry out a full classification of all non-periodic tilings of Z144." But the preprint that carries the full "Summary of the results for Z144" is "arXiv:0810.4338v1 [math.NT] 23 Oct 2008", footnoted "Date: October 2008"; 2009 is the journal issue (Crossref issued 2009-07). Amiot's 2009 paper in the same issue already calls it done ("n = 144 was completely and rather quickly solved by Matolcsi and Kolountzakis"). "Published in 2009" is true; "computed in 2009" is not what the sources show. Declined (listed, tier B MINOR): Tier B MINOR, decided from the checker's note: the arXiv preprint carrying the full Z144 summary is dated October 2008 and the journal issue is 2009, so 'computed in 2009' is a slip of precision, not a false fact a reader would carry away. The dek sentence holds no WRONG item, so it is listed, not edited. - CONFIRMED | Kolountzakis and Matolcsi settled 144 completely in 2009 and published their tile lists | https://api.crossref.org/works/10.1080/17459730903040899 | [dek-2009-journal-date] Crossref: Journal of Mathematics and Music 3(2), 85-97, issued 2009-07. Read as the publication year this holds; the preprint is October 2008 (see dek-2009). - CONFIRMED | tile lists at fourier.math.uoc.gr/~mk/tiling-computations/ | http://fourier.math.uoc.gr/~mk/tiling-computations/ | [tile-lists-published] Page (200) says "If you are only interested in the lists of non-periodic tiles please see the .txt files."; all four .txt files fetched today are byte-identical (cmp) to research/rhythmic-canons-144/km/*.txt. - MINOR | the paper that did it records which rhythm takes which partner in a single sentence | https://arxiv.org/pdf/0810.4338 | [km-single-sentence] KM print it as two sentences: "Sets in T1 tile with all sets in both T2 and S2. Sets in S1 only tile with sets in T2." Content is exactly as the page says; the count of sentences is not (the same wording recurs at md line 37 and index.html line 121, "in one sentence"). Declined (listed, tier B MINOR): Tier B MINOR: KM state the pairing in two consecutive short sentences ('Sets in T1 tile with all sets in both T2 and S2. Sets in S1 only tile with sets in T2.'); the content the page reports is exact. Loose wording in a sentence with no WRONG item: listed. - MINOR | that every summary table since has dropped, keeping only the two counts: 162 rhythms against 60 partners | https://arxiv.org/pdf/1304.6609 Table 1; amiottiljmm.pdf Figure 6; https://theses.hal.science/tel-01338353v2 | [summary-tables-dropped] Neither table read prints the KM sentence (searched "only tile with": 0 in Jedrzejewski and Amiot 2009, 1 in KM as control). But neither keeps "only the two counts: 162 ... against 60": Jedrzejewski's Table 1 prints the rows "12 12 6 60" and "12 12 162 12", which carry the pairing (the page itself says "Jedrzejewski's Table 1 does carry the pairing"), and Amiot 2009 prints "156 / +6" against "48 / +12". "Every summary table since" is a universal bounded by three documents; one more checked here (Caure's 2016 thesis) has no n = 144 table at all (its text contains "144" only as a page number). Declined (listed, tier B MINOR): Tier B MINOR: neither summary table prints the KM sentence (checker searched both), so 'dropped' holds; 'keeping only the two counts: 162 against 60' overstates, since Jedrzejewski's rows carry the pairing, as the page's own body says. Loose wording, no WRONG item in the sentence: listed. - CONFIRMED | 162 rhythms against 60 partners | https://arxiv.org/pdf/0810.4338 | [dek-162-60] "There are sets T1 and S1 containing respectively 6 and 156 non-periodic tiles of cardinality 12 (type A)... there are sets T2 and S2 containing respectively 12 and 48": 6 + 156 = 162, 12 + 48 = 60. - OBSERVED | one block of that pairing is empty and it is the largest one: 156 of the rhythms refuse 48 of the partners outright | recreated: node km-verify.mjs (scratch copy at 4fcb19ad5e) | [dek-empty-block] Printed "S1 x S2: 156 x 48 -> 0 canons (0 of 7488 pairs tile)"; blocks are 72, 288, 1872 and 7488 cells, so the empty one is the largest. - OBSERVED | Re-deriving every rhythm's complete partner list by exhaustive search fixes the incidence, and with it a total none of those sources states, 30,312. | recreated: node km-verify.mjs; node complements-check.mjs | [dek-total-30312] "TOTAL Vuza canons of Z_144, counted as pairs of translation classes: 30312" and "no tile has a partner the published lists do not contain, and no listed partner is missing." - CONFIRMED | That number is not in the sources this layer works from. | https://arxiv.org/pdf/0810.4338, research/rhythmic-canons-144/km/ (KM's four .txt files), https://arxiv.org/pdf/1304.6609, amiottiljmm.pdf (pdftotext) | [no-total-stated] Regex search for 30312, 35640, 30384, 28080, 9720, 7488 with optional separators: 0 hits in all; positive control 8640 found in KM (2), Jedrzejewski (1), Amiot 2009 (1). Amiot's PNM 2011 Figure 10 could not be read (closed access), so that one of the four is taken on the author's search. - WRONG | Every way of doing this was worked out in 2009, but the rhythms and the entry patterns were published as two separate lists, with no note of which goes with which. | https://arxiv.org/pdf/0810.4338; http://fourier.math.uoc.gr/~mk/tiling-computations/3-4-8--2-9-16.txt | [plain-two-lists] The 2009 publication did note it: the paper says "Sets in T1 tile with all sets in both T2 and S2. Sets in S1 only tile with sets in T2.", and the data file opens "For the notations T1, S1, T2, S2 consult the paper." The plain summary contradicts both the source and the page's own body ("the paper that did it records which rhythm takes which partner"). Fixed 2026-10-18 by the assay line. - CONFIRMED | That last clause is the hard one, and it is why the smallest cycle where this is possible at all is 72 beats. | https://arxiv.org/pdf/0810.4338 | [smallest-72] Section 1.1: "The smallest cyclic group that is not good is Z72." - CONFIRMED | the smallest cycle where this is possible at all is 72 beats | https://arxiv.org/pdf/1304.6609 | [smallest-72-jed] "computed the first "Vuza canon" that appears for N = 72, and the next one occurs at N = 108, 120, 144, 168, etc." (across the page 1/2 break). - WRONG | At 144 there are, it turns out, exactly 30,312 ways to do it. | https://arxiv.org/pdf/0810.4338; recreated: node km-verify.mjs | [exactly-30312] The sentence counts ways for "twelve drummers" each entering once on a 144-beat cycle. Only two of the four cases have twelve voices: KM give the first as "36 non-periodic tiles of cardinality 24 (type A), and a set T2 containing 6 non-periodic tiles of cardinality 6" and the third as "8640 non-periodic tiles of cardinality 24 ... 3 non-periodic tiles of cardinality 6"; km-verify prints "T1: 36 sets of size 24 / T2: 6 sets of size 6" and "T1: 8640 sets of size 24 / T2: 3 sets of size 6". Those 216 + 25,920 = 26,136 canons have 6 or 24 voices. Twelve drummers: 2,232 + 1,944 = 4,176. 30,312 is the count of all Vuza canons of Z144, any number of voices. Fixed 2026-10-18 by the assay line. - CONFIRMED | Amiot and Jedrzejewski both reprint the counts. | https://ar5iv.labs.arxiv.org/html/1304.6609; amiottiljmm.pdf | [reprint-counts] Jedrzejewski: "The results of Fripertinger (Fripertinger, 2001), Amiot (Amiot, 2011) and Kolountzakis and Matolcsi (Kolountzakis and Matolcsi, 2009) depicted in the first table are well-known."; Amiot 2009 Figure 6 prints 36/6, 8640/3, 156 +6 / 48 +12, 324/6. - CONFIRMED | for one of the four cases the two columns do not multiply, because most of the pairs do not tile | research/rhythmic-canons-144/km-verify.mjs (recreated: node km-verify.mjs) | [columns-dont-multiply] In that case 2,232 of 162 x 60 = 9,720 pairs tile, so 7,488 (77 percent) do not; scoped here to the one case, so true. - CONFIRMED | The case in question is the one Kolountzakis and Matolcsi themselves call the most interesting. | https://arxiv.org/pdf/0810.4338 | [most-interesting] "The partition {{3, 4, 8}, {2, 9, 16}} is the most interesting." - CONFIRMED | It has 162 rhythms and 60 partners. | https://arxiv.org/pdf/0810.4338 | [it-has-162-60] As dek-162-60. - CONFIRMED | Their prose records the pairing in a single sentence: six of the rhythms take all sixty partners, and the other 156 take only twelve. | https://arxiv.org/pdf/0810.4338 | [six-sixty-156-twelve] Content exact: T1 (6) tile with T2 and S2 (12 + 48 = 60); S1 (156) only with T2 (12). The one-sentence point is logged once, under km-single-sentence. - CONFIRMED | The summary tables downstream keep the counts and drop the sentence. | https://arxiv.org/pdf/1304.6609; amiottiljmm.pdf | [tables-keep-counts] Both print counts; neither contains the sentence ("only tile with" and "tile with all sets": 0 hits each, 1 each in KM as control). - UNVERIFIABLE | Amiot's Figure 10 prints the two sides as sums, "156 + 6" against "48 + 12" | tried: doi.org/10.1353/pnm.2011.0018, muse.jhu.edu/article/778070 (captcha), OpenAlex (oa_status closed, no repository copy), one WebSearch (no copy) | [amiot-fig10-a] The cited PNM 2011 Figure 10 was not readable. The same author's JMM 2009 Figure 6 (amiottiljmm.pdf, p. 13) prints exactly "156" over "+6" in "number of A's"; the record's excerpt is from that figure, not the cited one. - UNVERIFIABLE | Amiot's Figure 10 prints the two sides as sums, "156 + 6" against "48 + 12" | as amiot-fig10-a | [amiot-fig10-b] JMM 2009 Figure 6 prints "48" over "+12" in "number of B's"; PNM Figure 10 unread. - UNVERIFIABLE | aligned so that the natural reading pairs 156 with 48 | as amiot-fig10-a | [aligned-reading] In JMM 2009 Figure 6 the n = 144 row puts "156" and "48" on one line and "+6" "+12" on the next, and the first-listed type in each column is S1's ({3,4,6,8,12,24,48,72}) and S2's ({2,9,16,18,44}, a misprint for 144), so the aligned reading is indeed S1 x S2. Figure 10 itself unread. - OBSERVED | That is precisely the block in which nothing tiles: 7,488 pairs, every one of them a failure, and reading the table that way inflates the total to 35,640. | recreated: node km-verify.mjs | [block-7488-35640] "35640 table read as aligned"; by hand 156 x 48 + 6 x 12 + 216 + 25,920 + 1,944 = 35,640. - CONFIRMED | Jedrzejewski's Table 1 does carry the pairing, split across two rows, but the rows overlap on 72 canons, so adding them gives 30,384. | https://arxiv.org/pdf/1304.6609 | [jed-two-rows-30384] Rows for 144: "3 8640", "6 36", "6 60", "162 12", "324 6"; the paper says "The number of Vuza canons found is the product #S.#R"; 25,920 + 216 + 360 + 1,944 + 1,944 = 30,384, and 6 x 12 = 72 sits in both 6 x 60 and 162 x 12. - OBSERVED | For each of the 177 tiles on the cheaper side of each case we enumerated, by exact cover, the complete list of aperiodic partners | recreated: node complements-check.mjs | [enumerated-177] Tiles asked 6 + 3 + 162 + 6 = 177; last line "no tile has a partner the published lists do not contain, and no listed partner is missing." - OBSERVED | each of the six rhythms in the first family has exactly 60 partners, each of the 156 in the second has exactly 12, and the 7,488 remaining pairs have no tiling under any translation. | recreated: node complements-check.mjs; memory/log.d/2026-08-17T0520Z-which-goes-with-which.md | [sixty-twelve-7488] "each of the 6 tiles in T1 has exactly 60 aperiodic partners: 48 in S2 + 12 in T2" and "each of the 156 tiles in S1 has exactly 12 aperiodic partners: 12 in T2". - OBSERVED | the 36 and the 8,640 partners come out of the search rather than out of the file: those cases are re-derived entirely from their nine rhythms | recreated: node complements-check.mjs | [six-onset-nine] "each of the 6 tiles in T2 has exactly 36 aperiodic partners" and "each of the 3 tiles in T2 has exactly 8640 aperiodic partners"; 6 + 3 = 9. - MINOR | settled by an exhaustive sweep of all 464,306,843 six-element subsets of the cycle | research/rhythmic-canons-144/census144.c line 285 | [sweep-464m] The code says "brute: every d-subset containing 0." 464,306,843 = C(143,5), the six-element subsets containing 0; the cycle has C(144,6) = 11,143,364,232 six-element subsets. Exhaustive up to translation, which is what the question needs, but the number is mislabelled (same at index.html line 224). Declined (listed, tier B MINOR): Tier B MINOR: census144.c 'brute: every d-subset containing 0'; 464,306,843 = C(143,5), the six-element subsets containing 0, which is exhaustive up to translation; the label 'of the cycle' is imprecise, not a false result. No WRONG item in the sentence: listed. - OBSERVED | That was left running when this went up and nothing here rests on it; the lab's verifier asserts its result once folded in and skips loudly until then, so one command says which. | memory/log.d/2026-08-17T0520Z-which-goes-with-which.md; recreated: node verify.mjs | [sweep-unfinished] Log: "that sweep did not finish in this session." verify.mjs today: "(family I skipped: the n=144 six-onset sweep has not been folded in. It has never been run to the end." - CONFIRMED | The other two cases have twelve elements on both sides, where the honest statement is that we verified rather than reproduced: their exhaustiveness is Kolountzakis and Matolcsi's | research/rhythmic-canons-144/README.md; recreated: node km-verify.mjs | [verified-not-reproduced] Both remaining partitions are 12 x 12 ("T1: 6 sets of size 12 ... S2: 48 sets of size 12"; "T1: 6 sets of size 12 / T2: 324 sets of size 12"); README: "Their exhaustiveness is theirs." - OBSERVED | every family is closed under the 48 multipliers coprime to 144, which map canons to canons and would break an orbit if a tile were missing | recreated: node km-verify.mjs | [affine-closure] Four "affine closure HOLDS" lines, 2016 + 414864 + 10656 + 15840 = 443,376 images; phi(144) = 48 recomputed. - WRONG | all ten cyclotomic types their paper prints come back identical when recomputed by exact integer arithmetic | https://arxiv.org/pdf/0810.4338; research/rhythmic-canons-144/verify.mjs lines 76-90 | [ten-types-lead] KM's summary prints fourteen, all distinct: 3 for {2,8,9,16}/{3,4} (Phi{144,72,24,18,16,9,8,2}, Phi{144,72,18,16,9,8,2}, Phi{48,36,24,12,6,4,3}), 3 for {4,9,16}/{2,3,8}, 4 for {2,4,9,16}/{3,8}, 4 for {3,4,8}/{2,9,16}. verify.mjs FILES asserts the same 14 (2+1, 3+1, 1+1+1+1, 1+2) and family C passes, so "come back identical" holds for all fourteen; "ten" is wrong. Fixed 2026-10-18 by the assay line. - WRONG | all ten cyclotomic types their paper prints | https://arxiv.org/pdf/0810.4338 | [km-prints-types] As ten-types-lead: the paper prints 14 types for the four surviving partitions (plus five classes C1 to C5 for a discarded partition in its Step 3 example). Fixed 2026-10-18 by the assay line. - CONFIRMED | At 72 the single case is 3 rhythms by 6 partners and completely filled, so its counts multiply to 18. | https://arxiv.org/pdf/1304.6609; amiottiljmm.pdf | [fact-72-3x6] Jedrzejewski Table 1 row "72 2 2 3 3 2 6 12 3 6"; Amiot 2009 Figure 6 row 72: "6" A's, "3" B's; Fripertinger's length-72 compendium has every cell linked (see added line). - CONFIRMED | At 108 Fripertinger published the compendium as a drawn grid rather than a pair of counts, 252 rows by 3 columns with a midi file and a score in every cell where a canon exists | http://www.mathe2.uni-bayreuth.de/frib/canons/canon_4.html | [frip-108-grid] Byte-identical to the repo copy; "The first column desribes the inner rhythms, the first row the outer rhythms."; three outer rhythms, 504 data rows (252 x cycle and complete versions), each cell "mid : pdf"; 1512 .mid and 1512 .pdf links; RCMC108_0.pdf is a two-page score headed "108 - Kanon H. Fripertinger". - CONFIRMED | All 756 of its cells are filled. | http://www.mathe2.uni-bayreuth.de/frib/canons/canon_4.html (parsed here); recreated: node n108.mjs | [frip-756-filled] All 504 rows carry three .mid links; n108.mjs: "his grid: 756 of 756 cells filled -> COMPLETE". - OBSERVED | enumerating every aperiodic complement of each of his three rhythms returns exactly his 252, set for set. | recreated: node n108.mjs | [frip-252-set-for-set] "rhythm 0: 252 aperiodic partners by exact cover (7776 complements in all), 0 missing, 0 unlisted", same for rhythms 1 and 2. - CONFIRMED | 120 is also a Vuza order, it sits between 108 and 144, and nothing here touches it. | https://arxiv.org/pdf/1304.6609 | [order-120] "the next one occurs at N = 108, 120, 144, 168"; Table 1 has two rows for 120; KM also list N = 120 as a target. "Nothing here touches it" is about this layer, and the lab has no n = 120 output. - CONFIRMED | The published data files are Mathematica text exports hard-wrapped near column 78 | http://fourier.math.uoc.gr/~mk/tiling-computations/2-4-9-16--3-8.txt | [wrap-col78] Recounted: largest file max line 80, most lines 77 to 80; the other three max 70, 72, 70. The files use Mathematica list syntax and sit beside Mathematica 6 notebooks; the site does not itself call them exports. - CONFIRMED | in the largest of them the wrap falls inside a number four times: a line ends `1` and the next begins `29`, which is the single element 129 | http://fourier.math.uoc.gr/~mk/tiling-computations/2-4-9-16--3-8.txt | [wrap-inside-number] Exactly four digit-to-digit wraps, at lines 2652 ("127 , 1" then "29 , 135}"), 7932, 10566, 11882; none in the other three files. - CONFIRMED | Split on commas and trim, and one tile silently acquires an extra onset and stays plausible. | research/rhythmic-canons-144/km-verify.mjs | [split-comma-trim] Follows from the wrap at line 2652: "1" and "29" become two elements, so a 24-set gets 25. - CONFIRMED | The fix is to delete whitespace before splitting and then insist every set in a family have the same cardinality, which is what the verifier does. | research/rhythmic-canons-144/km-verify.mjs | [fix-delete-whitespace] parseFamilies: "g[1].replace(/\s+/g, '')" and a "sets of mixed cardinality" throw. - MINOR | The complete classification was computed in 2009, and the paper that did it says in one sentence | https://arxiv.org/pdf/0810.4338 | [html-classification-2009] index.html line 121. Same two near-misses as dek-2009 (preprint dated October 2008) and km-single-sentence (two sentences). Declined (listed, tier B MINOR): Tier B MINOR, the HTML dek copy of K1 and K2 (preprint dated October 2008; two short sentences, not one). Content correct; no WRONG item in this sentence: listed. - CONFIRMED | If every rhythm took every partner there would be 9,720 canons here | public/strata/which-goes-with-which/data.json (arithmetic over its counts) | [html-9720] 162 x 60 = 9,720. - OBSERVED | Every cell was decided by enumerating, from scratch, the complete list of aperiodic partners each rhythm has. | recreated: node complements-check.mjs | [html-every-cell] In this case the 162 rhythms (T1 and S1) were each asked; their lists decide all 9,720 cells. - OBSERVED | Every published set was re-tested for aperiodicity and for having a size that divides 144. All pass. | recreated: node km-verify.mjs | [html-aperiodic-size] "no problems: every published set is aperiodic, of a size dividing 144, and free of duplicate translation classes." - WRONG | was recomputed by exact integer polynomial arithmetic and matched against the ten types their paper prints. All ten match. | https://arxiv.org/pdf/0810.4338; research/rhythmic-canons-144/verify.mjs | [html-cyclo-ten] index.html line 220. Fourteen types printed and fourteen asserted (verify.mjs family C passes); see ten-types-lead. Fixed 2026-10-18 by the assay line. - OBSERVED | Each family was checked to be closed under the 48 multipliers coprime to 144 ... 443,376 images tested, none escaped. | recreated: node km-verify.mjs | [html-closed-443376] Sum of the four printed image counts is 443,376. - OBSERVED | For each of the 177 tiles on the cheaper side of each case, the complete list of aperiodic partners was re-derived by exact cover. No tile has a partner the published lists lack, and no listed partner is missing. | recreated: node complements-check.mjs | [html-177] As enumerated-177. - OBSERVED | re-parses Fripertinger's compendium, re-derives each rhythm's partner list by exact cover, and re-tests every canon his grid claims. | recreated: node n108.mjs | [html-n108-machinery] "every one of the 756 canons his table claims re-tested as a factorisation: 756". - OBSERVED | returns exactly his 252, set for set, and all 756 canons his table claims re-test as genuine factorisations. So 108 multiplies too, to 756. | recreated: node n108.mjs | [html-108-756] "the two counts DO multiply here: 3 x 252 = 756". - OBSERVED | It was left running when this went up and this page claims nothing from it. | memory/log.d/2026-08-17T0520Z-which-goes-with-which.md | [html-sweep-claims-nothing] Log: "that sweep did not finish in this session"; nothing on the page uses its result. - CONFIRMED | Somebody should check 120. The lab has the command. | research/rhythmic-canons-144/README.md | [html-lab-has-command] README usage "./census144 brute " and the log's "census144 120 one "; no n = 120 line is written out, but the generic command covers it. A stranger cannot reach the lab (see the last finding). - WRONG | the search that would do it is the one Kolountzakis and Matolcsi report as taking days | https://arxiv.org/pdf/0810.4338; amiottiljmm.pdf | [html-only-days] index.html line 510. KM's "several days" is about a different partition, one they never searched: "We have been able to execute the steps above for all partitions listed in Step 2 except for {{2, 4, 8, 16}, {3, 9}}. In that case ... the numerical search would take several days." The twelve-against-twelve partitions are among those they did execute, and Amiot 2009 says "n = 144 was completely and rather quickly solved" with computer time that "does not exceed a few hours on a personal computer". KM report no timing for the 12 x 12 searches. Fixed 2026-10-18 by the assay line. - CONFIRMED | There are four cyclotomic cases in all at 144 beats, and the other three are solid: every rhythm takes every partner, so there the counts do multiply. | https://arxiv.org/pdf/0810.4338 | [html-other-three-solid] Each of the other three summaries says "Each tile in T1 tiles with each in T2"; "The other partitions listed in Step 2 do not produce non-periodic tilings." - CONFIRMED | the same shape at 72 beats, where the single case is 3 rhythms by 6 partners and completely filled | https://arxiv.org/pdf/1304.6609 | [html-72-no-warning] As fact-72-3x6. - CONFIRMED | None of the three sources this layer works from states a total for 144 beats | https://arxiv.org/pdf/0810.4338 and research/rhythmic-canons-144/km/, https://arxiv.org/pdf/1304.6609, Amiot 2009 PDF | [html-totals-note] As no-total-stated; Amiot PNM Figure 10 unread, taken on the author's search. - OBSERVED | Together the four hold 30,312 Vuza canons of a 144-beat cycle, counted as pairs of translation classes, which is the convention under which 72 beats holds 18. | recreated: node km-verify.mjs; public/strata/which-goes-with-which/data.json | [built-rows] data.json "canons":30312 (live copy byte-identical); 216 + 25,920 + 2,232 + 1,944 = 30,312. - CONFIRMED | 28,080 from the three solid cases, plus 6x60 + 156x12 = 2,232 from the fourth. | public/strata/which-goes-with-which/data.json (arithmetic over its counts) | [built-measured-row] 216 + 25,920 + 1,944 = 28,080; 360 + 1,872 = 2,232. - CONFIRMED | Jedrzejewski's Table 1 gives this case as two rows, 6x60 and 162x12. | https://arxiv.org/pdf/1304.6609 | [built-jed-rows] "4 2 3 3 2 12 12 6 60" and "4 2 3 3 2 12 12 162 12". - UNVERIFIABLE | Amiot's Figure 10 prints "156 +6" against "48 +12". | as amiot-fig10-a | [built-amiot-print] Built at load time from data.json counts; matches Amiot's JMM 2009 Figure 6, cited Figure 10 unread. - CONFIRMED | They overlap on the 6x12 = 72 canons where a T1 rhythm meets a T2 partner, so adding them counts those twice. | public/strata/which-goes-with-which/data.json (arithmetic over its counts) | [built-overlap-72] 6 x 60 + 162 x 12 = 2,304 = 2,232 + 72. - CONFIRMED | M. N. Kolountzakis and M. Matolcsi, Algorithms for translational tiling, Journal of Mathematics and Music 3(2), 2009, 85-97 (arXiv:0810.4338) | https://api.crossref.org/works/10.1080/17459730903040899 | [cite-km] Title, authors, volume 3, issue 2, pages 85-97, 2009 all match; arXiv abs page has the same title, v1 only. - CONFIRMED | E. Amiot, Structures, Algorithms and Algebraic Tools for Rhythmic Canons, Perspectives of New Music 49(2), 2011, 93-142, Figure 10. | https://api.crossref.org/works/10.1353/pnm.2011.0018 | [cite-amiot] Title (with a comma after "Algorithms"), journal, 49(2), 93-142, 2011 match. The tool's "DOI MISMATCH (does not name Emmanuel Amiot)" is an artefact: Crossref stores {"family": "Emmanuel Amiot"} with no given name. Figure 10 itself unread. - CONFIRMED | F. Jedrzejewski, On the enumeration of Vuza canons, arXiv:1304.6609, Table 1. | https://arxiv.org/pdf/1304.6609 | [cite-jed] PDF v2 header "ON THE ENUMERATION OF VUZA CANONS"; arXiv listing title is "Enumeration of Vuza Canons"; Table 1 "Classification of Vuza Canons (up to N = 168)". - CONFIRMED | E. Coven and A. Meyerowitz, Tiling the integers with translates of one finite set, Journal of Algebra 212, 1999, 161-174. | https://api.crossref.org/works/10.1006/jabr.1998.7628 | [cite-coven] Crossref: Ethan M. Coven, Aaron Meyerowitz, J. Algebra 212(1), 161-174, 1999. - CONFIRMED | research/rhythmic-canons-144/ | git ls-tree 4fcb19ad5e research/rhythmic-canons-144 | [lab-path] Exists in the repository (not reachable by a reader; see findings). - CONFIRMED | Companion layer: No Beat Twice | src/content/strata/no-beat-twice.md | [companion-no-beat-twice] Exists; draws a rhythm and decides whether it tiles. - CONFIRMED | This layer's check reads the bytes this page serves and asserts against them. | research/rhythmic-canons-144/verify.mjs line 36 | [placard-bytes] verify.mjs reads public/strata/which-goes-with-which/data.json, which the page serves (live copy byte-identical). It does not read the page's sentences, so prose figures such as "ten" are not asserted. - MINOR | (added) Every table that has summarised it since keeps the two counts and drops the sentence. | https://arxiv.org/pdf/1304.6609 | index.html line 123. Same as summary-tables-dropped: Jedrzejewski's table keeps four numbers in two rows ("6 60", "162 12") that carry the pairing, as the page's own totals table says. Declined (listed, tier B MINOR): Tier B MINOR, the HTML copy of K3 (index.html line 123): the tables do drop the sentence; 'keeps the two counts' understates Jedrzejewski's two rows. Listed. - MINOR | (added) Most of the pairings turn out not to work, and this page shows which ones do. | research/rhythmic-canons-144/km-verify.mjs (recreated: node km-verify.mjs) | md line 9 (plain) and index.html line 111 (aw-plain meta). Unscoped, it reads as all pairings: across the four cases 7,488 of 37,800 class pairs fail (19.8 percent). Only within the fourth case is it most (7,488 of 9,720). Declined (listed, tier B MINOR): Tier B MINOR: 'Most of the pairings turn out not to work' is true within the fourth case (7,488 of 9,720) and loose unscoped (19.8 percent overall). It is its own sentence in the plain, not the sentence fixed for K4: listed. - CONFIRMED | (added) neither the rhythm nor the entry pattern ever repeats itself, and you have what the literature calls a Vuza canon | https://arxiv.org/pdf/1304.6609 | index.html lead. Definition 1: "A Vuza Canon (S, R) of G is a rhythmic canon G=S⊕R where neither S, nor R is periodic." - CONFIRMED | (added) At both orders where somebody drew the grid, the two counts were enough. | http://www.mathe2.uni-bayreuth.de/frib/canons/canon_3.html; canon_4.html | Fripertinger drew both: canon_3.html is "The complete RCMC-canon compendium of length 72" with every data row fully linked; KM: "H. Fripertinger listed out all non-periodic tilings of Z72 and Z108". - OBSERVED | (added) Everything downstream of it, the incidence and the total, is ours and reruns in about a minute. | recreated: time node km-verify.mjs; time node complements-check.mjs | 6.2 s and 61.7 s wall on this machine (verify.mjs, the full check, 113 s). - MINOR | (added) Everything done to them here is ours, and it is the part you can rerun | curl https://github.com/artwasteland/Artificial-Wasteland; curl https://artwaste.land/research/rhythmic-canons-144/ | index.html lines 215-216, and "whoever reads this can find out in one command" (line 224). The lab is named only as a code path with no link; the repository answers 404 anonymously (control: github.com/artwasteland answers 200) and artwaste.land/research/rhythmic-canons-144/ is 404 (control: artwaste.land/strata/which-goes-with-which/sets-p2.json is 200). A stranger cannot rerun anything. Reader-side, likely corpus-wide. Declined (listed, tier B MINOR): Tier B MINOR: the lab is named as a repository path and the repository is not public, so 'the part you can rerun' overstates what a stranger can do; it is a statement about the page's own checking, and per the checker likely corpus-wide. Listed. ## Findings - [fixed] plain-two-lists (WRONG): the plain summary (frontmatter plain and the aw-plain meta) said the rhythms and entry patterns were published "with no note of which goes with which". KM's paper says "Sets in T1 tile with all sets in both T2 and S2 . Sets in S1 only tile with sets in T2 ." It now says the paper says which goes with which and the summary tables since left that note out; "worked out in 2009" in the same sentence became "published in 2009". Corrected line added. - [fixed] exactly-30312 (WRONG): md body "At 144 there are, it turns out, exactly 30,312 ways to do it", after "Hand twelve drummers one rhythm". Cases p1 and p2 have 6 or 24 voices (KM: cardinalities 24 and 6); it now says 30,312 once the number of drummers is left free, 4,176 of them with twelve (data.json totals.byPartition p3 2,232 + p4 1,944). Corrected line added. - [fixed] ten-types-lead (WRONG): md body "all ten cyclotomic types their paper prints" now "all fourteen"; KM's Summary prints 14 (2+1, 1+2, 3+1, 1+1+1+1) and verify.mjs asserts 14. Corrected line added (one line for the three copies). - [fixed] km-prints-types (WRONG): the same statement as ten-types-lead; fixed with it, and the lab README and the verify.mjs header comment also now say fourteen. - [fixed] html-cyclo-ten (WRONG): index.html "matched against the ten types their paper prints. All ten match." now "fourteen ... All fourteen match."; covered by the same Corrected line. - [fixed] html-only-days (WRONG): the honesty note said the twelve-against-twelve search "is the one Kolountzakis and Matolcsi report as taking days"; KM's "several days" is about {{2, 4, 8, 16}, {3, 9}}, which they settled by argument. It now gives the lab's own reason (README, types.mjs: at |A| = 12 the type constrains almost nothing); README "What this lab does not establish" fixed to match. Corrected line added. - [declined] computed-in-2009 (MINOR, K1): "The complete classification of the aperiodic rhythmic canons of a 144-beat cycle was computed in 2009". Listed, not edited, under the tier B rule: Tier B MINOR, decided from the checker's note: the arXiv preprint carrying the full Z144 summary is dated October 2008 and the journal issue is 2009, so 'computed in 2009' is a slip of precision, not a false fact a reader would carry away. The dek sentence holds no WRONG item, so it is listed, not edited. - [declined] one-sentence (MINOR, K2): "the paper that did it records which rhythm takes which partner in a single sentence". Listed, not edited, under the tier B rule: Tier B MINOR: KM state the pairing in two consecutive short sentences ('Sets in T1 tile with all sets in both T2 and S2. Sets in S1 only tile with sets in T2.'); the content the page reports is exact. Loose wording in a sentence with no WRONG item: listed. - [declined] summary-tables-only-two-counts (MINOR, K3): "that every summary table since has dropped, keeping only the two counts: 162 rhythms against 60 partners". Listed, not edited, under the tier B rule: Tier B MINOR: neither summary table prints the KM sentence (checker searched both), so 'dropped' holds; 'keeping only the two counts: 162 against 60' overstates, since Jedrzejewski's rows carry the pairing, as the page's own body says. Loose wording, no WRONG item in the sentence: listed. - [declined] sweep-subsets-of-cycle (MINOR, K6): "settled by an exhaustive sweep of all 464,306,843 six-element subsets of the cycle". Listed, not edited, under the tier B rule: Tier B MINOR: census144.c 'brute: every d-subset containing 0'; 464,306,843 = C(143,5), the six-element subsets containing 0, which is exhaustive up to translation; the label 'of the cycle' is imprecise, not a false result. No WRONG item in the sentence: listed. - [declined] computed-in-2009 (MINOR, K9): "The complete classification was computed in 2009, and the paper that did it says in one sentence". Listed, not edited, under the tier B rule: Tier B MINOR, the HTML dek copy of K1 and K2 (preprint dated October 2008; two short sentences, not one). Content correct; no WRONG item in this sentence: listed. - [declined] summary-tables-only-two-counts (MINOR, K12): "(added) Every table that has summarised it since keeps the two counts and drops the sentence.". Listed, not edited, under the tier B rule: Tier B MINOR, the HTML copy of K3 (index.html line 123): the tables do drop the sentence; 'keeps the two counts' understates Jedrzejewski's two rows. Listed. - [declined] most-pairings-unscoped (MINOR, K13): "(added) Most of the pairings turn out not to work, and this page shows which ones do.". Listed, not edited, under the tier B rule: Tier B MINOR: 'Most of the pairings turn out not to work' is true within the fourth case (7,488 of 9,720) and loose unscoped (19.8 percent overall). It is its own sentence in the plain, not the sentence fixed for K4: listed. - [declined] reader-cannot-rerun (MINOR, K14): "(added) Everything done to them here is ours, and it is the part you can rerun". Listed, not edited, under the tier B rule: Tier B MINOR: the lab is named as a repository path and the repository is not public, so 'the part you can rerun' overstates what a stranger can do; it is a statement about the page's own checking, and per the checker likely corpus-wide. Listed.