2017-18 Prizm Mosaic

• G-001 NAME DIVERGENCE, THE HEADLINE FINDING OF THIS BUILD, AND A BYTE-FOR-BYTE MATCHER KEYED ON EITHER SOURCE STRING WILL MISS THIS RELEASE. Four spellings are in play for one product and the one written into product_name is used by NONE of the sources. (1) Panini America itself, in its own announcement, prints "2017-18 Mosaic Prizm Basketball" - see "Tomorrow at 10 a.m. CST, Panini America will make company history with the release of 2017-18 Mosaic Prizm Basketball" and the post title "The Panini America Quality Control Gallery: 2017-18 Mosaic Prizm Basketball (72 Pics)" at https://blog.paniniamerica.net/2018/03/13/the-panini-america-quality-control-gallery-2017-18-mosaic-prizm-basketball-72-pics/. (2) cardboardconnection prints "2017-18 Panini Mosaic Prizm Basketball Cards" as its H1 at https://www.cardboardconnection.com/2017-18-panini-mosaic-prizm-basketball-cards. (3) sportscardchecklist prints "2017-18 Panini Mosaic Prizm" and "2017-18 Panini Mosaic Prizm Autographs (33 Cards)" at https://www.sportscardchecklist.com/sport-basketball/year-2017-18/index-panini-mosaic-prizm/trading-card-checklists-and-product-information. (4) tcdb prints plain "2017-18 Panini Mosaic" with no second word at all, on the parent heading at https://www.tcdb.com/ViewSet.cfm/sid/169909/2017-18-Panini-Mosaic and on every one of its twelve child headings. So the word order is Mosaic Prizm on three sources, absent on the fourth, and the catalogue string is Prizm Mosaic. product_name is nonetheless written byte-identical to reference/brands.csv ("Panini Prizm Mosaic Basketball", brand id brd_01M2B8M6BJKF4BVCZ1A77FRZ28, line_prefix prizmmosaic) because the owner decided on 2026-09-27 to fold 2017-18 into the prizmmosaic line rather than mint a third brand id - specs/2017-18_SCOPE_MEMO.md section 8a. NOTHING WAS MINTED AND brands.csv WAS NOT TOUCHED BY THIS BUILD. CONSUMER IMPACT: a resolver matching a marketplace listing, an editorial page or a tcdb heading for this release against product_name will fail on all four source strings; it needs the aliases "Panini Mosaic Prizm", "Mosaic Prizm" and "Panini Mosaic" to reach this release row. This also refutes the 2018-19 memo section 3.1 claim that tcdb "reverses" the order as an indexing quirk: on this season tcdb renders neither order, and the manufacturer itself renders Mosaic Prizm. The order tracks the SEASON - Mosaic Prizm 2017-18, Prizm Mosaic 2018-19, Mosaic 2019-20 onward - and that is now confirmed against a first-party source rather than an editorial one. Fifth occurrence noted for completeness: groupbreakchecklists uses BOTH orders at once, a page titled "2017-18 Mosaic Prizm Basketball" under a URL path segment reading 2017-18-prizm-mosaic-basketball.
• G-002 RELEASE DATE: THE ONLY EDITORIAL DATE FIELD IS QUALIFIED, AND THE DATE IS WRITTEN ANYWAY ON TWO FURTHER SOURCES. cardboardconnection's field reads "Release Date: 03/14/2018 (first online sale date)" (https://www.cardboardconnection.com/2017-18-panini-mosaic-prizm-basketball-cards) - a qualified field, not a plain release date. It is NOT an "Estimated Release Date", so the extraction spec section 3.2 refusal rule does not apply. The date was written as 2018-03-14 for three reasons, each checkable: (a) tcdb's parent set page renders an UNQUALIFIED "Release Date: March 14, 2018" (https://www.tcdb.com/ViewSet.cfm/sid/169909/2017-18-Panini-Mosaic), a structurally independent source that applies no qualifier at all; (b) Panini America, posting on March 13, 2018, writes "Tomorrow at 10 a.m. CST, Panini America will make company history with the release of 2017-18 Mosaic Prizm Basketball" and "iCollectPanini.com tomorrow beginning at 10 a.m. CST" (https://blog.paniniamerica.net/2018/03/13/the-panini-america-quality-control-gallery-2017-18-mosaic-prizm-basketball-72-pics/) - first-party, naming a fixed on-sale time one day ahead, which is the opposite of a forecast; (c) the qualifier names the same event here BECAUSE the product has only one channel. Panini states it was "offered exclusively and entirely on iCollectPanini.com", so there is no wide or hobby release for a first online sale date to precede. THE RESIDUAL RISK, STATED: for a product that is not online-exclusive, a first online sale date and a release date can differ, and this reasoning must not be carried to any other release. cardboardconnection also writes "for $100 per box starting March 14", a second in-body rendering of the same day.
• G-003 CAVEAT CAV-036 EXTENDED, TWELFTH 2017-18 PARENT AND THE WORTHLESS SHAPE AGAIN. tcdb's Inserts.cfm on this parent (https://www.tcdb.com/Inserts.cfm/sid/169909/2017-18-Panini-Mosaic) carries a Notes column in which ALL TWELVE rows hold an image-completion percentage ("0.0% of images added" through "40.0% of images added"), there are ZERO blank cells and ZERO serial statements in any format, prose or bare token. So the both-directions split test cannot be run on this page at all and the blank-annotation signal is unavailable rather than negative. This is the state CAV-036 records as the single most common one across eleven 2017-18 parents, now measured on a twelfth. Note the SHAPE DIFFERS FROM THE ADJACENT SEASON: prizmmosaic-bkb-2018-19 recorded that its parent has NO Notes column whatsoever; this parent HAS the column and it is useless, which is a different failure and would defeat a check written to detect the 2018-19 one. Every print run in parallels.csv consequently rests on a per-sid row-level token sweep enumerating the DISTINCT token set of all 100 or 33 rows of each of the twelve tiers, with nothing sampled.
• G-004 THE AUTOGRAPHS SET IS SERIAL NUMBERED WITH A PER-CARD VARIABLE RUN AND sets.csv CANNOT CARRY THE FIGURE. There is no print_run column in sets.csv and the extraction spec section 2.2 forbids putting the figure in card_count. A full enumeration of all 33 rows returns a DISTINCT TOKEN SET OF EXACTLY TWO VALUES, SN99 and SN49, so no single figure is true of the set. cardboardconnection states only a ceiling, in words: "With a print run of 99 or less, the Autograph set features legends like Kobe Bryant and key rookies such as Donovan Mitchell." (https://www.cardboardconnection.com/2017-18-panini-mosaic-prizm-basketball-cards) - and a ceiling is not a print run. THE PER-ROW SPLIT IS CONTESTED AND IS NOT WRITTEN ANYWHERE AS A VALUE: https://www.tcdb.com/Checklist.cfm/sid/169918/2017-18-Panini-Mosaic-Autographs?sort=0 reports 28 rows SN99 and 5 rows SN49, naming the SN49 rows as A-CP3 Chris Paul, A-KBR Kobe Bryant, A-KDR Kevin Durant, A-RGM Reggie Miller and A-SON Shaquille O'Neal; https://www.tcdb.com/ViewSet.cfm/sid/169918/2017-18-Panini-Mosaic-Autographs reports 27 SN99 and 6 SN49 and adds A-MJS Magic Johnson to that list - while quoting A-MJS's own token as "AU, SN99" in the same response, a self-contradiction inside one fetch of exactly the shape the 2017-18 scope memo section 7.2 records. What BOTH agree on and what is safe: the distinct token set is exactly {SN99, SN49}, the set is 33 cards, and every row carries AU. Whether Magic Johnson is /49 or /99 is unresolved. This is the extraction spec section 2.3 expressiveness gap in its sets.csv form.
• G-005 base_print_run=unnumbered carries base_print_run_basis=assumed, not sourced, and deliberately so. No source found states in words that the base cards carry no serial numbering. The evidence is a positive discriminative absence: no serial token renders on any of the 100 base rows on three endpoints (https://www.tcdb.com/Checklist.cfm/sid/169909/2017-18-Panini-Mosaic?PageIndex=1 rows 1-81, https://www.tcdb.com/Gallery.cfm/sid/169909/2017-18-Panini-Mosaic?PageIndex=9 rows 81-90, https://www.tcdb.com/Gallery.cfm/sid/169909/2017-18-Panini-Mosaic?PageIndex=10&sort=0 rows 91-100) while the same renderers print SN1, SN10, SN25 and SN99 on all 100 rows of four of this release's own parallel tiers. Panini America's box-break line "will deliver nine Prizms and either an autograph or a numbered Prizm" (https://blog.paniniamerica.net/2018/03/13/the-panini-america-quality-control-gallery-2017-18-mosaic-prizm-basketball-72-pics/) implies the same by contrast, since only the parallels are called numbered. Strong indirect evidence is still assumed, per extraction spec section 2.1.
• G-006 SECTION 5.3 CHECK RUN AND RECORDED SO ITS RESULT IS EVIDENCED RATHER THAN ASSUMED: there is no set filed inside the parent checklist as a numbered block past base_set_size, so the exclusion rule had nothing to exclude. The parent states "Total Cards: 100" (https://www.tcdb.com/ViewSet.cfm/sid/169909/2017-18-Panini-Mosaic); the enumeration runs a contiguous 1-100 with no holes and no b-suffixed or duplicate numbers rendered anywhere; Gallery page 11 renders ZERO TILES (https://www.tcdb.com/Gallery.cfm/sid/169909/2017-18-Panini-Mosaic?PageIndex=11) against a pagination control showing exactly ten pages; and the header agrees with the row count, so no Total Cards unit question arises. The Autographs set carries letter codes with an "A-" prefix, not numbers continuing past 100, so it cannot be such a block.
• G-007 Errors.cfm ON THE PARENT RETURNED SILENCE AND THAT SILENCE IS BEING TREATED AS EVIDENCE OF NOTHING, per caveat CAV-020. https://www.tcdb.com/Errors.cfm/sid/169909/2017-18-Panini-Mosaic returns "No errors or variations have been added for this set yet." NO CLAIM IS MADE THAT THIS RELEASE HAS NO VARIATION SET. What survives is what rests on independent enumeration: base_set_size=100, a contiguous 1-100 base run, and no numbered block past it. A variation layer would render as b-suffixed rows, which Gallery.cfm strips and which a contiguous base-run enumeration is not designed to surface, so the possibility is unsupported rather than disproved. tcdb also renders no VAR or UER token on any of the 100 base rows or any of the 1,100 parallel-tier rows swept for this build, which is a weaker and differently-sourced signal pointing the same way.
• G-008 rookie_flag RESTS ON A SINGLE PRIMARY SOURCE, WITH ONE CORROBORATION THAT MAY SHARE ITS UPSTREAM. 29 of the 100 base rows carry a tcdb RC designation and 71 do not (https://www.tcdb.com/Checklist.cfm/sid/169909/2017-18-Panini-Mosaic?PageIndex=1, https://www.tcdb.com/Gallery.cfm/sid/169909/2017-18-Panini-Mosaic?PageIndex=9, https://www.tcdb.com/Gallery.cfm/sid/169909/2017-18-Panini-Mosaic?PageIndex=10&sort=0); no editorial source consulted states rookie status card by card. A free corroboration was run against this project's own already-gated data/prizm-bkb-2017-18/base_cards.csv, which shares all 100 player names: rookie_flag agrees on 100 of 100 of them, zero disagreements. That is a real check and it is NOT independent evidence, because that release may have taken its RC designations from the same tcdb upstream. Recorded as single-sourced.
• G-009 CHANNEL AXIS CONFLICT, and channel_sku is deliberately EMPTY. Every parallel row carries channel=panini-direct with channel_basis=sourced-exclusive on a product-scope first-party exclusivity statement - "offered exclusively and entirely on iCollectPanini.com", "the official online store of Panini America" (https://blog.paniniamerica.net/2018/03/13/the-panini-america-quality-control-gallery-2017-18-mosaic-prizm-basketball-72-pics/) - corroborated by cardboardconnection "sees the popular brand return but this time as an online exclusive" and "albeit direct to consumers via the online Panini America Store for $100 per box starting March 14" (https://www.cardboardconnection.com/2017-18-panini-mosaic-prizm-basketball-cards). That satisfies CAV-010: the sources state EXCLUSIVITY, not mere presence online, so sourced-exclusive rather than sourced-presence is earned. THE CONFLICT: the same sources describe the box in hobby terms - cardboardconnection writes "each 2017-18 Panini Mosaic Prizm Basketball hobby box averages one autograph or one numbered base parallel" and "Product Configuration: 10 cards per pack, 3 packs per box, 12 boxes per case", and Panini America repeats the same configuration. channel conflates trade channel, box SKU and region and holds one value, and reference/channel_taxonomy.csv classes hobby as a trade-channel rather than a box type, so there is no box-type name here to put in channel_sku - no blaster, mega, jumbo or hanger is named. channel_sku is therefore empty and the hobby axis is recorded here rather than forced into either column. A consumer asking "was this a hobby product?" cannot answer it from the channel column.
• G-010 THE FOUR UNNUMBERED TIERS ARE RECORDED source_confidence=single, NOT multi, AND THE REASON IS A RULE ABOUT OMISSION. Blue, Green, Orange and Red carry print_run=unnumbered on a full 100-row token sweep showing an empty distinct token set, corroborated in appearance by the fact that BOTH editorial sources annotate exactly four of eight tiers and leave these four unannotated - Panini America "Red, Blue, Orange, Purple (numbered to 99), Camo (numbered to 25), Gold (numbered to 10), Black 1/1s and Green." (https://blog.paniniamerica.net/2018/03/13/the-panini-america-quality-control-gallery-2017-18-mosaic-prizm-basketball-72-pics/) and cardboardconnection "Green, Red, Blue, Orange, Purple #/99, Camo #/25, Gold #/10, Black 1/1." (https://www.cardboardconnection.com/2017-18-panini-mosaic-prizm-basketball-cards). Per extraction spec section 2.1 an editorial checklist OMITTING a figure is not a statement of fact, so that agreement does not make a second source for the VALUE unnumbered, and these rows are single-sourced to the tcdb sweep. The four numbered tiers are multi. A SEPARATE AMBIGUITY THIS RESOLVED, recorded because it would otherwise be a silent coin-flip: Panini's list reads "Red, Blue, Orange, Purple (numbered to 99)" and the parenthetical could attach to all four names or to Purple alone. The token sweep settles it - Purple renders SN99 on all 100 rows while Red, Blue and Orange render nothing on any of theirs - and cardboardconnection's independent comma placement reads the same way. Had the four-name reading been taken, three tiers would have been written /99 falsely.
• G-011 FILING DIVERGENCE AGAINST tcdb'S OWN CATEGORISATION, REPRODUCED FROM THE ADJACENT SEASON. tcdb lists Autographs Black (sid 169919), Autographs Camo (169920) and Autographs Gold (169921) under its "Insert Sets (4)" heading alongside the Autographs set itself, rather than under "Parallel Sets (8)" (https://www.tcdb.com/Inserts.cfm/sid/169909/2017-18-Panini-Mosaic). They are written here as set-scope rows in parallels.csv with scope=Autographs, because each renders "Total Cards: 33" matching the Autographs set exactly and each carries the same 33 "A-" card codes as that set, which is a parallel and not a distinct chase set. sportscardchecklist files them the same way this build does, as tiers of the Autographs checklist (https://www.sportscardchecklist.com/sport-basketball/year-2017-18/index-panini-mosaic-prizm/trading-card-checklists-and-product-information). prizmmosaic-bkb-2018-19 recorded the identical divergence for its own three autograph tiers, so this is a property of how tcdb files this line rather than a one-off.
• G-012 Beckett was unreachable for this release, one more consecutive confirmation. https://www.beckett.com/news/2017-18-panini-mosaic-prizm-basketball/ returns HTTP 302 to https://maintenance.beckett.com/ and was not followed. A web search surfaced that URL under the result title "2017-18 Panini Mosaic Prizm Basketball Checklist Takes Release Online", which would be a fourth independent rendering of the Mosaic Prizm word order if the page could be read - but a search-engine result title is not the page and it is NOT cited as evidence anywhere in this release. The 2017-18 scope memo section 7.6 records Beckett as confirmed gone for this season; this is the same finding on a 2017-18 product URL.
• G-013 No source found states a licensing grantor for this product. license_status=licensed (NBA), licensed=yes and league=NBA are written on the catalogue-wide convention of the weaker sourceable claim. Nobody on this project has ever sourced a grantor, and licensed (NBA/NBPA) is deliberately NOT written. Team names render on every base row, which evidences NBA team marks and nothing about the players association.
• G-014 A RELEASE EXISTS BELOW THIS LINE'S FLOOR AND WAS NOT CHASED, AND IT IS THE FOURTH SUCH FINDING OF THIS SEASON. tcdb's 2016-17 index carries a "2016-17 Panini Mosaic" entry filed under Promo, recorded in specs/2017-18_SCOPE_MEMO.md section 6 item 3 and explicitly left unaddressed by the owner decision in section 8a, which folded only 2017-18 into prizmmosaic and claimed no floor. This build was instructed not to chase it and did not fetch it; existence is asserted on the scope memo alone, not on anything this build read. cardboardconnection's own framing supports something below 2017-18: "2017-18 Panini Mosaic Prizm Basketball sees the popular brand return but this time as an online exclusive" (https://www.cardboardconnection.com/2017-18-panini-mosaic-prizm-basketball-cards) - a brand that RETURNS had a prior appearance, so 2017-18 is not this line's floor. The season's other three floor findings are recorded in the scope memo section 4 addendum for revolution, optic and crownroyale.
• G-015 CONTRADICTS THE 2017-18 SCOPE MEMO SECTION 7.1, WHICH MEASURED THE OPPOSITE ON THREE PARENTS. That section states "The parent sid sits nowhere near its children on all three" at distances of 7,158, 2,355 and 4,225 sids, and instructs that Inserts.cfm rather than outward sweeping is the enumeration. ON THIS PARENT THE CHILDREN ARE IMMEDIATELY ADJACENT: parent 169909, children 169910 through 169921 with no gap and no interleaving, and BOTH boundaries proved foreign - 169908 renders "2002 Rittenhouse James Bond Die Another Day - East Coast National Promos" (https://www.tcdb.com/ViewSet.cfm/sid/169908/boundary-probe-below) and 169922 renders "2003 Donruss Champions - National Sports Collectors Convention Wrapper Redemption" (https://www.tcdb.com/ViewSet.cfm/sid/169922/boundary-probe-above). The memo's advice still holds operationally - Inserts.cfm was used and delivered all twelve rows in one fetch with no truncation and no interpolated sid, all twelve confirmed against their own page headings - but its stated finding is now one parent out of four in the other direction, and it should not be read as a regime. The sid arithmetic reconciles exactly: twelve child sids equal eight base-scope parallels plus one chase set plus three set-scope parallels, and equal sportscardchecklist's thirteen entries minus the base set.
• G-016 CONTRADICTS OUR OWN ADJACENT-SEASON RELEASE ON A COLUMN THE DATABASE DOES NOT POLICE. data/prizmmosaic-bkb-2018-19/parallels.csv leaves channel_confidence EMPTY on all eleven of its rows while carrying a known channel of panini-direct on every one. This build writes channel_confidence =multi on all eleven of its rows, per caveat CAV-038, which records that channel_confidence takes the same three values as source_confidence (multi, single, UNKNOWN), carries NO database CHECK, and is therefore invisible when wrong or absent. The two seasons of this line now disagree about whether the column is populated, and a consumer filtering on it gets this release and not its sibling. The value multi is written because two sources state the exclusivity and each carries a fact the other lacks: Panini America names the store (iCollectPanini.com) and cardboardconnection names the price ($100 per box), which two pages copying one another wholesale would not do. THE RESIDUAL RISK, STATED PLAINLY: cardboardconnection plausibly descends from Panini's own announcement, so these may be one lineage and the honest floor for this value could be single.
• G-017 THE BASE PARALLEL LADDER REVERSES AGAINST THE ADJACENT SEASON ON TWO TIERS, AND THE SIBLING SEASON WAS NOT USED AS EVIDENCE ANYWHERE IN THIS BUILD. prizmmosaic-bkb-2018-19 records Blue, Green and Red unnumbered, Orange 99, Purple 49, Camo 25, Gold 10, Black 1. This release measures Blue, Green, Red AND ORANGE unnumbered, PURPLE 99, Camo 25, Gold 10, Black 1 - so Orange loses its serial numbering entirely and Purple moves from 49 to 99, while the eight tier NAMES are identical across both seasons. An agent inheriting the sibling ladder would have written /99 on Orange and /49 on Purple, two false sourced integers on 200 cards. The suspicious-uniformity fabrication check fires on the identical eight-name ladder and is dismissed by exactly this divergence: a template fill would have reproduced the runs as well as the names. All eight runs come from full 100-row token sweeps on this season's own sids.
• G-018 NOTES FOR THE CHASE CHECKLIST PASS, RECORDED HERE BECAUSE THIS BUILD SAW THEM AND insert_cards.csv IS OUT OF ITS SCOPE. (1) THE PREFIX CHANGED BETWEEN THE TWO SEASONS OF THIS LINE. The Autographs codes are "A-" on 2017-18 (A-ADV Anthony Davis, A-KAT Karl-Anthony Towns, A-SON Shaquille O'Neal) against "MO-" on prizmmosaic-bkb-2018-19. Do not carry either forward; re-derive. (2) The set size also moved, 33 on 2017-18 against 31 on 2018-19. (3) All three colour tiers carry the SAME 33 codes as the base Autographs page, in the same order, so the parallel-page test is available in both directions on this set. (4) Every row of all four autograph sids carries the literal token AU and none renders EXCH, so no redemption or exchange placeholder was seen; that is an observation about what rendered, not a claim that none exists. Full row-level source: https://www.tcdb.com/Checklist.cfm/sid/169918/2017-18-Panini-Mosaic-Autographs?sort=0. (5) The base Autographs page is a mixed print-run page - see the gap on the SN99 and SN49 split.
• G-019 COPY CHECK AND SECTION 5.2 IDENTITY CHECK, RUN AGAINST OUR OWN DATA AT NO FETCH COST, WITH A REAL FINDING. The 100-card base roster of this release is a STRICT SUBSET OF THE 300-CARD BASE ROSTER OF data/prizm-bkb-2017-18: all 100 player-name strings appear there, exact-string, with zero exceptions. It is NOT the section 5.2 shared-checklist collision, because the NUMBERING is entirely different - only one of the 100 cards happens to sit on the same number in both releases (81 Andrew Wiggins), which is chance at that rate. So these are different physical cards on a shared roster, not one checklist under two release ids, and nothing was deduplicated or double-counted. FOUR TEAM DISAGREEMENTS BETWEEN THE TWO RELEASES, not resolved by this build and not errors on their face, since the two products shipped four months apart across a trade deadline: this release renders Dwyane Wade as Cleveland Cavaliers where prizm-bkb-2017-18 renders Chicago Bulls; Eric Bledsoe as Milwaukee Bucks against Phoenix Suns; Enes Kanter as New York Knicks against Oklahoma City Thunder; and Carmelo Anthony as Oklahoma City Thunder against New York Knicks. Each pair is a real 2017-18 transaction and the direction is consistent with this release being the later one. Against the other two Mosaic-family releases the roster overlap is partial and no copy signal fires: 60 of 100 names shared with prizmmosaic-bkb-2018-19 and 81 of 100 with the 300-card mosaic-bkb-2019-20.
• G-020 FABRICATION-CHECK REGISTER, recorded so a later pass can see which checks were RUN rather than assumed. COPY CHECK: run, fired, explained - see the gap on the roster subset; the eight parallel tiers genuinely ARE full 100-card base parallels and each was confirmed by its own sid rendering "Total Cards: 100". TEMPORAL IMPOSSIBILITY: run and passed - all 29 rookie rows are 2017 draftees, all 71 veterans are 2017-18 NBA players, and the four team divergences above run in the direction the release dates predict. SUSPICIOUS UNIFORMITY: fired on the identical eight-name ladder shared with 2018-19 and dismissed because two of the eight runs differ. SET-NAME CROSS-CHECK: the two set names and all eleven tier names were read on three independent sources (tcdb, cardboardconnection, sportscardchecklist) and agree. SOURCE CALIBRATION: sportscardchecklist was tested only on questions that could be checked inside tcdb and passed all thirteen - its (100 Cards) and (33 Cards) figures match tcdb's Total Cards headers AND this build's own independent row enumerations on every entry (https://www.sportscardchecklist.com/sport-basketball/year-2017-18/index-panini-mosaic-prizm/trading-card-checklists-and-product-information); it was not relied on for any print run or card number. TWO SUMMARISING-FETCH DEFECTS CAUGHT AND NOT PROPAGATED: a Camo fetch asserted "80 rows are displayed" above a list of 79, and the Autographs ViewSet fetch contradicted itself inside one response on Magic Johnson's token. Row counts in this release are the build's own counts of enumerated items, never a fetch's stated total. NARROW RE-FETCH: run on the two load-bearing editorial strings - Panini's parallel list was re-asked under a second URL specifically for its comma placement, and its exclusivity clause re-asked as a bounded fragment, because the first reading had truncated both.

If you know a source that answers one of these, tell us — name the question's label and where to look.