Hypotheses and their falsification tests

Written 2026-08-22, before any site was scored. The mission requires that a major hypothesis carry, in advance, what would support it, what would contradict it, what field observation would strengthen it, and what would weaken it. Writing them after scoring would let each test be shaped to protect a result already produced.

This file is also a prerequisite for the ranking. ranking/ranking.md cannot score a site's D - what a visit would settle - until every site has a test here, because otherwise D measures how much attention a site has already had rather than what it can decide. Levelling that is the whole point.

Adequacy is settled. The search-adequacy threshold landed 2026-08-22 - two gravel bars, 45 minutes each, low water, and a chert control that must fire; one adequate negative moves confidence down one step. Full rules in ../identification/fieldwork.md. Where a test below says "adequate", it means adequate by that file, and a negative may now move a rating.

H1 - The bridge: the manganese chert and the gem jasper are the same material

Statement. The brightly colored chert Davis described in 1917 around grant lot 16 (PJ-0004, PJ-0005, PJ-0006) is the same rock as the orbicular jasper of the Watsonville Road deposit (PJ-0011, PJ-0016), 5-6 km south.

Standing: HYPOTHESIS. Not established, and the register is built to keep it that way - the lineage field holds the two evidence bodies apart and ranking/ranking.md Cap 1 stops cross-lineage evidence raising a score above 1.

What supports it - Both are Franciscan chert in a contact setting (tier 1 mapping). - Davis 1917: "the chert is brightly colored as a result of the contact action of these igneous rocks" (tier 2), with a mechanism. - Bulletin 152 p.269 states the chert/jasper equivalence outright (tier 2). - Proximity, 5-6 km. This is the weakest support and must never be counted as evidence.

What contradicts it - The two literatures never cite each other, across 1917-1966. - No manganese source mentions orbs, orbicular structure, or gem use. - No gem source mentions manganese. - Different sub-watersheds - Little Llagas (301) and Upper Llagas (302), which data/hydrography.md shows do not exchange material above Lower Llagas. Whatever the two places share, it is not a transport path. - Different mapped settings: serpentinite / melange at the prospects, an fc chert body at the deposit.

Field observation that would STRENGTHEN it. Orbicular structure - orbs, not merely color or banding - found in chert at the lot 16 district. A single unambiguous orbicular specimen in place there would do it.

Field observation that would WEAKEN it. An adequate search of exposed chert at the lot 16 district finding colored and manganese-stained chert but no orbicular structure anywhere. Adequacy: Class A (../identification/fieldwork.md) - all accessible chert exposure within the located lot 16 boundary walked in daylight, with the position resolved to reach or better first. "Not exposed" is not "not present" here more than anywhere: these are 1917 prospect pits, and Trask records the Morgan Hill workings as already abandoned.

Desk tests that could settle it without a visit - OFR 75-303 (ILL 237598767) naming the gem commodity behind the I-909 dots. - ~~Lot 16's boundary courses~~ RUN 2026-08-22, partly answered. Lot 16's centroid is in a band at 37.132-37.136 N, -121.683 to -121.690 W; both prior estimates excluded; MRDS 10140986 excluded, reversing the register's duplicate-record position. Not survey-grade - the traverse could not be closed. Book 116 of Maps p. 10 is in hand and very likely fixes the anchor. See references/lot-16-survey.md.

A shape finding that bears directly on this hypothesis. Lot 16 is a headwater block - bounded by two named 1892 drainages, its south corner on a water divide, its north line on the Las Uvas Rancho boundary. So whatever Davis examined in 1917 was at or near the top of a drainage, not in a valley bottom - and Davis's "area west of Mr. Kelley's house" points to the western part of the lot, toward the divide.

Why that matters to H1. The gem locality is a creek-bank outcrop in a valley floor; this is a headwater block on a divide. Same rock type would still be possible - Franciscan chert occurs in both settings - but the two occurrences would be in different structural positions, which is a further way in which "same material" needs evidence rather than proximity. It neither supports nor refutes H1; it sharpens what the hypothesis is claiming.

Testability. - To strengthen: one unambiguous orbicular specimen in place in lot 16 chert - gated - key: the lot 16 anchor first (Book 116 of Maps p. 10, in hand, not yet plotted - references/lot-16-survey.md), then permission from whichever parcel owner that anchor identifies. Two keys in sequence; the second cannot be asked for until the first is turned. - To weaken: a Class A search of lot 16 chert finding color but no orbs - blocked - route: a route would need bedrock exposure across the lot that is not there to walk. These are 1917 prospect pits, abandoned by Trask's time, and "not exposed" is not "not present". Short of re-excavation there is no representative negative, so this hypothesis can only ever be confirmed, never refuted. That asymmetry is why it is retained as a constraint - the lineage field and ranking Cap 1 - rather than pursued as a question. - By document: OFR 75-303's manganese summary - in flight - ILL 237598767, due before 2026-11-30. Bears on it; does not settle it.

Tested by: PJ-0004, PJ-0005, PJ-0006, PJ-0015 (positive side); PJ-0016 (the comparison specimen).

H2 - The interpillow mechanism

Statement. Orbicular jasper here forms as hydrothermal chert in interpillow pelagic deposits within pillow basalt, and that mechanism operates in the Permanente terrane (fpv) where the gem occurrences sit, not only in the Marin Headlands terrane (fmv) where MF-2373 documents it.

Standing: HYPOTHESIS. MF-2373 is tier 1 and asserts the mechanism only for fmv. Its fpv description does not mention jasper. Nearest mapped fmv is 1.18-1.75 km from the prospects.

What supports it - MF-2373's fmv text names "hydrothermal chert (orbicular jasper)" with hematite in the assemblage - the project's only tier-1 statement of a mechanism, and orbicular is the poppy-jasper habit. - Interpillow chert is a general property of pillow basalt, not a terrane-specific one.

What contradicts it - No source asserts it for fpv. The extension is entirely this project's inference. - PJ-0001 sits on fpv with all three maps agreeing, and no chert is mapped there.

Observation that would STRENGTHEN it. Orbicular jasper found in interpillow position within mapped fpv - that is, a chert pod between recognizable pillow forms, not float.

Observation that would WEAKEN it. fpv pillow basalt exposures in this area examined and shown to lack interpillow chert entirely. This disproof condition was already recorded in data/geology.md and is carried here unchanged.

Testability. - To strengthen: orbicular jasper in interpillow position within mapped fpv - gated - key: a position for PJ-0001 or PJ-0002 better than the MRDS +/-1000 m, then permission from the owner of the parcel it lands in. Both are zone-extent and unresolved. - To weaken: fpv pillow exposures examined and shown to lack interpillow chert - gated - key: the same ground, the same two keys.

Tested by: PJ-0001 (on fpv), PJ-0002.

An observation added 2026-08-22. The two creek-adjacent hills in the deposit corridor whose relief matches the 1936 "probably 200 feet high" - 223 ft and 162 ft, both in Paradise Valley proper - are mapped fg, greenstone, which is altered basalt. The right rock class for this mechanism sits at the right target geometry beside the creek. Not confirmation: Dibblee's fg and MF-2373's fmv are different classifications and no interpillow chert is mapped at either hill. Recorded because it is the first time the mechanism and the 1936 geometry have pointed at the same ground. See data/geology.md.

H3 - El Toro is a propagation error

Statement. The attribution of poppy jasper to El Toro is an error that entered the record in 2006 and spread, and no material ever came from there.

Standing: STRONG, and this project's most confidently held negative.

What supports it - No source before 2004 attaches El Toro to poppy jasper (references/claim-genealogy.md). - The attribution traces to a single San Jose Mercury News article of 2006-11-12 - "back side of El Toro" - and the dealer pages that copied it. - El Toro maps as greenstone with silicified-limestone lenses, a different host rock from the fc varicolored chert at Watsonville Road. - Zero hits for "El Toro" across all seven CDMG volumes examined, controlled.

What would contradict it - Any pre-2004 source attaching El Toro to jasper. - Material with credible El Toro provenance whose thin section matches deposit material.

Observation that would WEAKEN it. Orbicular jasper found in place on El Toro. Note this would not merely weaken H3 - it would break the host-rock model, because the two localities map as different rocks.

Observation that would STRENGTHEN it. An adequate search of El Toro's accessible exposures finding no jasper. Weak: El Toro is largely private and the search would be unrepresentative.

The cheap decisive test is not a visit. A thin section of any specimen with a documented El Toro provenance, compared against deposit material. Note the standing prohibition: a dealer or show-floor specimen labeled "El Toro" documents a label, not a location, and cannot enter this chain.

Testability. - To strengthen: an adequate search of El Toro finding no jasper - blocked - route: a route would need enough of El Toro to make the search representative, and El Toro is largely private with no register site because no evidence ever put one there. The project does not intend to seek this; the standing negative rests on documents and needs no field time. - To weaken: orbicular jasper found in place on El Toro - gated - key: El Toro's landowners, unnamed, and a find nobody has ever reported. - By specimen: a thin section of a provenanced El Toro specimen against deposit material - blocked - route: a specimen with documented pre-2004 El Toro provenance. None is known to exist, and a dealer or show-floor label cannot enter this chain.

Tested by: no register site. This hypothesis has no field target, which is itself worth recording - El Toro is not in the register because nothing evidential ever put it there.

H4 - Downstream concentration below the outcrop

Statement. Material eroded from the outcrop (PJ-0016) concentrates in identifiable depositional traps downstream, in the Llagas reach through Paradise Valley.

Standing: MODERATE, and it is the project's own prediction rather than anyone's report.

What supports it - The 1936 account documents exactly this having happened: Pitts found "loose pebbles and small boulders in Llagas Creek bed" half a mile below the outcrop. The mechanism is attested, not merely modelled. - Pitts's own 1947 account describes downstream fining - "little pieces" low down, "large chunks" moving upstream - the normal signature of a local source. - The reach is below Chesbro Dam, so it was never sediment-starved by the 1955 impoundment.

What contradicts it - CJMG 1954 on Llagas Creek gravel composition: "arkosic sandstone, chert, shale, basalt, gabbro, and quartz." Jasper is not listed. This is the only contemporary description of what the creek actually carries, and it is a genuine negative - weakened, but not removed, by Bulletin 152's chert/jasper equivalence. - Pitts claims to have "hauled out fifty tons", and says "Not much of it exists today that is not badly fractured." The reach may simply have been cleaned out by 1947.

Observation that would STRENGTHEN it. Orbicular jasper float found in the Llagas channel between PJ-0016 and PJ-0014, or at the Watsonville Road crossing.

Geometry corrected 2026-08-22: that reach runs northwest to southeast through Paradise Valley, from about 37.0902/-121.6621 down to the bridge at 37.0849/-121.6534, roughly 1.1 km of channel. An earlier reading had it running north-south and would have sent a search across a hillside instead of along the creek.

Observation that would WEAKEN it. An adequate Class B search finding Franciscan chert but no orbicular jasper. Adequacy per ../identification/fieldwork.md: two separate bars, 45 min each, low water, chert control firing. Note the access reality: the lawfully reachable portion of this reach is currently the Watsonville Road right-of-way, which is Class C and can never produce a negative. Weakening H4 therefore requires permission first - the observation is defined, but it is not currently available.

Testability. - To strengthen: orbicular float in the Llagas channel between PJ-0016 and PJ-0014, or at the crossing - open at PJ-0014 - the Watsonville Road right-of-way is public and a Class C sightline can confirm a find. It can confirm; it cannot search. - To weaken: an adequate Class B search of the reach finding chert but no orbs - gated - key: permission for the PJ-0013 reach from the parcel owner of record (data/access.md), by the route in data/permission-route.md. So on current access this hypothesis can be confirmed but not refuted.

Tested by: PJ-0013, PJ-0014, PJ-0011.

H5 - Bowen's host rock: altered serpentine, not Franciscan chert

Statement. The host is altered serpentine, as Bowen 1962 states, rather than the Franciscan fc chert body this project has been building its model on.

Standing: OPEN CONFLICT, deliberately unresolved.

Bowen, Oliver E., Jr., 1962, Rocks and Minerals of the San Francisco Bay Region, UC Press - Bowen was a CDMG geologist: "The most famous California localities for orbicular jasper are Paradise Valley and the Coyote vicinity in Santa Clara County. There it occurs in veins and irregular masses in altered serpentine or in stream beds draining such terrain."

Why this hypothesis matters more than it looks: H5 and H1 are coupled.

If Bowen is right, the manganese prospects are in the right setting after all - PJ-0004 maps as serpentinized ultramafic and PJ-0005 sits on a serpentinite/melange contact (data/geology.md). Confirming an altered-serpentine host would therefore raise H1 substantially, because the two evidence bodies would share a host rock rather than merely a neighborhood. It would also mean this project has been reading Dibblee's fc chert body as the source on the strength of a coincidence of position.

One observation tests two hypotheses. That is exactly what D is meant to reward, and PJ-0016 is where it happens.

What supports Bowen - A CDMG geologist, in a UC Press book, stating a host rock with a habit - veins and irregular masses - which matches the 1936 account's vein 10 feet wide better than a bedded chert does. - The 1936 geometry is a vein through a hill, not a bed. Veins are more at home in serpentine than in bedded radiolarian chert.

What contradicts Bowen - Dibblee 1973 plate 4 maps an fc varicolored chert body in a small hill on the west side of Watsonville Road - the right description in the right place (tier 1). - Bowen names no specific locality within "Paradise Valley" and may be generalising from the Coyote side, where serpentine is better developed.

Observation that would STRENGTHEN Bowen. Host rock at the outcrop found to be serpentine or silica-carbonate altered ultramafic, with jasper in veins cutting it.

Observation that would WEAKEN Bowen. Host rock at the outcrop found to be bedded or massive radiolarian chert, with no serpentine in contact.

Testability. - To strengthen: host rock at the outcrop found to be serpentine or silica-carbonate rock with jasper in veins - gated - key: permission at PJ-0016 from the parcel owner of record (data/access.md). One competent look at the host rock runs both tests. - To weaken: host rock found to be bedded or massive radiolarian chert with no serpentine in contact - gated - key: the same look, the same permission.

Tested by: PJ-0016 - decisively, by a single competent look at the host rock.

Confirmed available at one location, 2026-08-22. Dibblee plate 4, georeferenced with a passing control, maps serpentine (sp) 225 m from PJ-0014 at the creek bend, and fc varicolored chert ~500 m from the same point on the south flank of the adjacent hill. Both competing host rocks are mapped within 500 m of one hill, so they are not geographically separable here and one look tests both. See data/geology.md.

H6 - "Paradise jasper" is a New Almaden name, not a Morgan Hill one

Statement. Symons 1940 is right and Shipley 1945 is wrong: paradise jasper originally denoted material from New Almaden.

Standing: OPEN CONFLICT, deliberately unresolved. Both sources are institutionally respectable and they contradict each other.

• Symons, Henry H., 1940, Rocks and Minerals v.15 no.2 - Curator of Museum, State Division of Mines: "South of San Jose near New Almaden... also from this same locality comes the so-called paradise jasper."
• Shipley 1945, GIA Dictionary of Gems and Gemology: "paradise jasper. A local trade name for a variegated red jasper from Morgan Hill, Calif." - carried verbatim into the USBM dictionary (Thrush 1968).

What this bears on. Nomenclature only - not the location of the deposit, which is established by four converging lineages independent of the name. Recorded because an unresolved conflict must not quietly disappear.

RE-DESCRIBED 2026-09-15, and the hypothesis has less support on BOTH sides than it looked. Two facts verified this day change the shape of it:

• Symons is on both sides of his own conflict. CJMG v.31 (1935) lists him as Statistician and Curator while the Division accessioned "PARADISE JASPER ... from 15 miles northwest of Hollister" - which computes in EPSG:26910 to Gilroy, 14.0 mi at 319 deg. The same curator wrote "near New Almaden" five years later. One man, two localities, and neither of them is Morgan Hill.
• Shipley 1945 may not be an independent voice at all. It is reported, though not verified here, to restate Dake 1938, which is separately established as the first source to attach "Morgan Hill" to this material. If so, the "Shipley side" of the conflict is Dake's claim repeated, and the USBM dictionary of 1968 is that same claim repeated a third time.

So this is no longer "a curator against a dictionary." It is one curator contradicting his own department's accession, against a dictionary that may be quoting a 1938 handbook. Three apparent evidence bodies collapse toward one, which is a loss of apparent support rather than a gain. See references/primary-sources.md.

Observation that would STRENGTHEN it. A pre-1940 source attaching paradise jasper to New Almaden independently of Symons - most likely the Gems & Gemology serial glossary.

Observation that would WEAKEN it, and the project already holds one. A pre-1940 source attaching the name to the Morgan Hill material. references/primary-sources.md records exactly that, and it had not been connected to this conflict until 2026-08-22:

Rocks and Minerals v.11 no.3 (March 1936), p.48, the magazine's Acknowledgments column: "Eyles, Wilfred C. ... Five beautiful specimens orbicular and paradise jaspers from near Gilroy"

That is four years before Symons, and it is attached to Wilfred C. Eyles - the man who found the outcrop, donating specimens from it. "Near Gilroy" is the Morgan Hill locality's own standing locator ("7 miles NW of Gilroy").

So the evidence currently leans against Symons, and H6's standing should be read as weak rather than balanced. The fork stays open anyway, for a reason worth stating: Symons was a state museum curator writing about specimens he presumably handled, and both localities producing material under the same trade name is entirely possible - trade names travel with material, not with geology. Ruling that out needs the glossary, not an argument.

What would settle it. REVISED 2026-09-08 - the document this hypothesis nominated has been read, and it cannot settle anything. Gems & Gemology was obtained free (GIA and archive.org; no ILL, and the "not digitized" note in pending-documents.md item 12 was simply wrong) and searched in full. paradise jasper = 0 across all 363 issues, 1934-2023, and paradise = 0 in the Jan 1934 - Winter 1940 serial-glossary window itself, under coverage, flattening and de-hyphenation controls, with jasper = 35 and GEMOLOGICAL GLOSSARY = 62 in that same window proving the search live. See references/primary-sources.md.

So Shipley did not take the entry from the glossary he cites, and this hypothesis has no nominated settling document. Candidates now, none strong:

• Identify Shipley's actual source. This is a new open question created by the negative - his stated basis does not contain his entry.
• A pre-1940 New Almaden attribution independent of Symons, from any source. That remains the STRENGTHEN condition; only its expected location has been removed.
• The 1937 cumulative index to The Mineralogist vols 1-4 (ILL 237598735, to be re-filed), which would name any article in that run using the term.

Standing is unchanged and still leans against Symons, on the March 1936 Rocks and Minerals acknowledgment. What changed is only that the cheap resolution is gone.

Field test: none. This is a document question and should never consume field time.

Testability. - 2026-09-08 both lines used to be gated on the Gems & Gemology 1934-40 run. That run has been read and the term is absent from it, so the gate is gone and neither line can name it any more. - To strengthen: a pre-1940 source attaching paradise jasper to New Almaden independently of Symons - blocked - route: would need Shipley's actual source identified, since his stated basis does not carry the entry, or a systematic sweep of the 1936-40 trade press starting with the Mineralogist vols 1-4 index (ILL 237598735, awaiting re-filing). - To weaken: a further pre-1940 source attaching the name to the Morgan Hill material - open by document - the Desert Magazine and Rocks and Minerals 1936-40 runs are freely available and unswept. The project already holds one such source (Rocks and Minerals, March 1936), so this line is corroborative rather than decisive.

Tested by: no register site.

H7 - A second occurrence outside the Pajaro basin, at New Almaden

RESTATED 2026-08-22. It previously read "east of the Pajaro/Coyote divide" and rested on the A. G. Swanson ranch and the Anderson Lake dot. Both of those supports failed, and a better one appeared at a different place. Full working: references/coyote-side.md.

Statement. Orbicular jasper occurs outside the Pajaro basin, as transported isolated boulders, in the New Almaden district / Alamitos Creek - HUC8 18050003 Coyote, 18.92 km from PJ-0016 with a drainage divide between.

Standing: MODERATE (was: hypothesis). Two independent tier-1/2 sources, one of them a federal monograph.

What supports it - USGS Professional Paper 360 (Bailey & Everhart 1964, tier 1): "'Orbicular jaspers' … are fairly common as isolated boulders" in the New Almaden district, with a two-stage formation mechanism and an explicit note that "similar rocks occurring a few miles east of the district in the Morgan Hill quadrangle are valued by lapidary enthusiasts." - Bowen 1962 (tier 2): "Paradise Valley and the Coyote vicinity" - naming two localities, independently of PP 360.

What failed, and this is why the hypothesis was restated rather than merely upgraded - The A. G. Swanson ranch is UNLOCATED. Controlled negative: Swanson = 0 across the whole of CJMG v.50 no.2 - the state's own county mineral inventory - against present controls magnesite 91, O'Connell 6, Cochrane 3, Morgan Hill 7. Also 0 across 455 issues of Desert Magazine (one hit, the known item), GNIS statewide, and TIGER. So the WEAKEN condition was not met either - the ranch is not on the Pajaro side; it is nowhere yet. - PJ-0003 now carries a dated NEGATIVE. The full 1951 sentence says the Anderson Dam stop collected magnesite and massive calcite - not jasper. The truncated version this project held concealed that. H7 no longer rests on PJ-0003 at all. - Swanson Canyon (GNIS 235876) is the county's only Swanson toponym and is a weak correlation: Pajaro / Upper Uvas, 13.2 km from Morgan Hill, 17.1 km from Anderson Dam against the item's "near-by", and not a single Franciscan polygon within 2 km.

What contradicts it - "A few miles east of the district" admits two readings and neither was established. - A divide-separated occurrence needs its own source; PP 360 supplies a mechanism but no bedrock body in the Llagas system.

Observation that would STRENGTHEN it. Orbicular jasper with documented provenance in the Alamitos Creek drainage - not at Anderson Lake, which is a different sub-basin and now carries a negative.

Observation that would WEAKEN it. An adequate search of accessible Alamitos Creek exposures finding none, or locating the Swanson ranch on the Pajaro side, which would collapse the 1950 trip into the main locality.

Testability. - To strengthen: orbicular jasper with documented provenance in the Alamitos Creek drainage - open at PJ-0018 - open public parkland on the district PP 360 describes. zone extent keeps it off the shortlist; it does not keep anyone off the ground. - To weaken: an adequate search of accessible Alamitos Creek exposures finding none - open at PJ-0018. The alternative - locating the Swanson ranch on the Pajaro side - is desk work with no bound on it.

Tested by: PJ-0018, Almaden Quicksilver County Park / Alamitos Creek, registered 2026-08-22 - open public parkland on the district PP 360 describes, mapping as fmv at the creek reach, the one unit whose own tier-1 description names orbicular jasper. It is the first register entry this hypothesis has ever had. PJ-0003 is no longer its test - a correction worth carrying, so the entry is not read as holding evidence it does not have. references/coyote-side.md recommends a New Almaden District entry citing PP 360, not MRDS (whose record for it contains zero jasper, chert or gem terms).

H8 - A second source in the Llagas headwaters, above Chesbro

Statement. Orbicular jasper occurs in the upper Llagas catchment above Chesbro Dam, from a source independent of the Watsonville Road deposit.

Standing: HYPOTHESIS, cheap to test, and attempted once - inadequately. The 2026-08-22 visit to PJ-0008 found no orbicular jasper but failed both coverage tests, so it is not an adequate negative and moved nothing. The test is still open; see Testability below.

Why it cannot be deposit material. PJ-0008 sits at lat 37.132-37.144; the dam is at ~37.121 and the creek flows south. Material there is upstream of the 1955 sediment trap and upstream of the deposit itself. There is no mechanism by which deposit material arrives. A confirmed find is therefore not a transport result - it is a second source.

What supports it - A firsthand collector report (tier 6): "at Rancho Cañada del Oro Open Space Reserve - walk along Longwall Canyon Trail and check the creek bed you have to cross - we have found some there." - The preserve sits in Franciscan greenstone (SGMC, tier 1) - greenstone is altered basalt, and MF-2373 attributes orbicular jasper to hydrothermal chert in interpillow deposits within pillow basalt. The right rock class is present. - The catchment above Chesbro is mapped Franciscan melange containing chert blocks "red, white or green" (OFR 98-795, tier 1).

What contradicts it - A single anonymous tier-6 report, with the trail name unverified. - No historical source places jasper above Chesbro. - "Orbicular jasper" in a collector report may be any patterned chert.

Observation that would STRENGTHEN it. Orbicular jasper observed and photographed in place in the creek bed at the trail crossing.

Observation that would WEAKEN it. An adequate Class B search of that crossing finding chert but no orbicular jasper. This is one of the few sites where a real negative is currently obtainable, because access is lawful.

Access makes this the cheapest test in the project. Free public open space, open daily, maintained trail, a 2024 bridge over Llagas Creek, 1.86 km of creek inside the boundary. Collecting is prohibited (OSA Reso 18-10), so it is observation and photographs - which is all this hypothesis needs.

Testability. - To strengthen: orbicular jasper observed and photographed in place at the trail crossing - open at PJ-0008 - free public open space; collecting prohibited, and observation is all this needs. - To weaken: an adequate Class B search of the crossing finding chert but no orbs - open at PJ-0008. - 2026-08-22 weaken attempted: one reach of about 25 m, about 90 minutes; the chert control fired and the low-water test was met, but both coverage tests failed, so it is NOT an adequate negative and PJ-0008's rating held at speculative. An adequate re-run is a second visit away - two separate bars, 45 minutes each. See register/PJ-0008-rancho-canada-del-oro-longwall-canyon.md.

Tested by: PJ-0008 directly. PJ-0009 is its downstream consequence: if H8 is true, the Chesbro reservoir margins and inflow deltas are where that material accumulates, since the dam traps everything the upper catchment sheds. If H8 is false, PJ-0009 has no source and its rating should fall.

H9 - The Lake Silveira quarry exposes buried Llagas alluvium

Statement. The flooded aggregate quarry at Atherton Way is an artificial exposure of Llagas Creek's own buried alluvium, and its spoil and faces record what the creek carried before the modern channel.

Standing: HYPOTHESIS. The site is in the register on a tier-6 report; this is the first statement of what it would actually test.

The three facts, joined here for the first time. They were held in three separate files: - PJ-0007 is a tier-6 collector report at "Atherton Way - Hidden Pond." - Lake Silveira is an 8.2-acre flooded 1970s aggregate quarry on Llagas Creek (data/hydrography.md). - PJ-0007's coordinate falls inside the Silveira Property - the 53.4-acre County Parks holding at Atherton Wy that CPAD wrongly classes Open Access and the county's own GIS records as closed/land bank (data/access.md). Verified by point-in-polygon 2026-08-22.

Why a quarry is worth more than another gravel bar. A modern bar shows what the creek is moving now. A quarry cuts down through the deposit, so it exposes material that arrived decades to millennia earlier - including any pulse from the historic deposit before Pitts removed "fifty tons" of it. It is the only place in the register offering a vertical section rather than a surface.

Position, established 2026-08-22 by the ordered channel walk: PJ-0007 is 2.81 km downstream of the float locality and 3.89 km downstream of the outcrop, on the same undammed reach below Chesbro. Transport is mechanically available - which was not previously established and had been assumed.

So this is also H4's farthest-downstream test. If material concentrates downstream of the outcrop, 3.9 km is where you find out whether it travels or fines out.

What contradicts it - Aggregate quarries process and remove their product. What remains is reject, and reject is selected against - jasper is hard and would report to the good fraction. - The pond is flooded, so faces may be underwater. - CJMG 1954 lists no jasper in Llagas gravel composition.

Observation that would STRENGTHEN it. Orbicular jasper in quarry spoil, tailings or an exposed face.

Observation that would WEAKEN it. An adequate Class B search of accessible spoil finding Franciscan chert but no orbicular jasper. That would be a strong negative for H4 generally, because a quarry samples far more of the deposit than a bar does.

Access: ask, and the owner is already known - Santa Clara County Parks. That is a better starting position than any private parcel in the corridor, and the county's own Sec. B14-55.1 provides a research and education specimen permit route. Do not enter - the holding is closed land bank, and CPAD's "Open Access" classification is a known error this project has already documented.

Testability. - To strengthen: orbicular jasper in quarry spoil, tailings or an exposed face - gated - key: a research and education specimen permit from Santa Clara County Parks under Sec. B14-55.1. The holder is known and the route is written; nobody has asked yet. - To weaken: an adequate Class B search of accessible spoil finding chert but no orbs - gated - key: the same permit.

Tested by: PJ-0007.

H10 - The manganese chert reaches Little Llagas Creek

Statement. If the lot 16 manganese chert is gem material (H1), it should occur as float in Little Llagas Creek below the prospects.

Standing: HYPOTHESIS. This is H1's downstream test, and it may be reachable when the prospects themselves are not.

Why it is worth separating from H1. The prospects are zone-extent, positionally disputed, and privately held. Little Llagas Creek is undammed and runs through the town of Morgan Hill (data/hydrography.md), passing roughly 1.1 km from both prospects. Testing H1 from downstream avoids needing to stand on a prospect nobody can currently locate to better than +/-1000 m.

What supports it - Little Llagas (HUC12 ...0301) does not pass through Chesbro - its transport path is undammed and continuous. data/hydrography.md calls this "the cleanest source-to-transport link available." - Davis 1917 describes brightly colored chert in place at the prospects.

What contradicts it - Channelisation. Little Llagas through Morgan Hill has had flood-protection works, which can strip gravel, bury it, or expose it. data/hydrography.md flags this as unresolved and not predictable from a map. - H1 itself is unproven, so a negative here is ambiguous - it weakens either H1 or this transport link, and does not say which.

Observation that would STRENGTHEN it. Orbicular jasper float in Little Llagas below the prospects.

Observation that would WEAKEN it. An adequate Class B search finding chert but no orbicular jasper. Read carefully: this weakens the conjunction of H1 and H10, not H1 alone.

Unresolved and needed first: whether any reach of Little Llagas Creek is publicly accessible, and whether the channelised sections retain natural gravel. Neither is known.

Testability. - To strengthen: orbicular float in Little Llagas Creek below the prospects - gated - key: ourselves, first - an access determination for the Little Llagas reaches below the prospects (data/access.md), which nobody has made - then whatever owner that names. The cheapest gate in the file, because the first key is desk work this project can do. - To weaken: an adequate Class B search finding chert but no orbs - gated - key: the same determination, plus whether the channelised reaches retain natural gravel at all.

Tested by: PJ-0010.

H11 - a source in the upper Paradise Valley reach, upstream of PJ-0016

Statement. Poppy jasper has a bedrock source in the Paradise Valley reach of Llagas Creek upstream of PJ-0016, between it and Chesbro Dam - ground no register entry covered before 2026-08-22.

Standing: HYPOTHESIS, and it is new. It exists because three unrelated lines of work converged on the same 1 km of creek without being aimed there.

What supports it - and note that none of the three was looking for it

1. A trap that needs a supply. PJ-0017 is the only reach in 5.87 km of channel where three different depositional mechanisms coincide within 330 m - an inside bend, the corridor's largest terrace remnant, and its strongest gradient break. It sits 1,024 m UPSTREAM of PJ-0016, and a trap upstream of a source cannot collect that source's material. Either something feeds it from further up, or it is empty.
2. Hills of the right size, made of the right rock. The only two creek-adjacent hills whose relief matches the 1936 "small hill that is probably 200 feet high" sit 578 m and 910 m from the trap cluster - and both are fg greenstone, altered basalt, the rock class H2's interpillow mechanism requires. Established on the georeferenced Dibblee plate.
3. PJ-0016's own uncertainty permits it. That entry carries +/-1 km of along-channel uncertainty and has already moved once. Its position is a projection of a projection: a 1936 textual bearing, projected onto the channel, then walked 806 m upstream.

What contradicts it - No source of any tier places material in this reach. Not one. material_basis for PJ-0017 is none, and that is the honest state. - The 1936 account puts the outcrop "about half a mile" above the float, which - if PJ-0011 is right - lands at PJ-0016, not a kilometer further up. - H4's downstream-concentration model and this hypothesis compete for the same evidence. Jasper found between PJ-0016 and PJ-0014 supports H4; jasper found at PJ-0017 supports this. They are not both economically the same story, and the register should not let a find at one be read as support for both.

Observation that would STRENGTHEN it. Orbicular jasper in the PJ-0017 bar or terrace gravel - which, being upstream of every source the project has named, could only have come from something further up.

Observation that would WEAKEN it. An adequate Class B search of PJ-0017 finding Franciscan chert but no orbicular jasper. That is a strong negative and worth more than most, because this is the best-formed trap in the corridor: a good trap with no jasper says the supply is not there, rather than that the search was unlucky.

A desk test that could move it cheaply. Aerial imagery over the PJ-0017 reach - does it hold exposed gravel at all, or has it been scoured, buried or gravel-mined? A trap on paper that is a concrete channel on the ground costs a field day to discover.

Testability. - To strengthen: orbicular jasper in the PJ-0017 bar or terrace gravel - gated - key: permission for the PJ-0017 reach from the parcel owner of record (data/access.md). - To weaken: an adequate Class B search of PJ-0017 finding chert but no orbs - gated - key: the same permission. - By desk: aerial imagery over the PJ-0017 reach - open - does it hold exposed gravel at all? A scoured or concreted channel kills this for nothing, before anyone asks for a key.

Tested by: PJ-0017 directly. PJ-0016 indirectly - if the outcrop turns out to be upstream of the cluster rather than below it, this hypothesis collapses into a correction of that entry rather than a separate finding.

H12 - the three I-909 gemstone dots represent poppy jasper

Statement. The three gemstone symbols USGS Map I-909 plots in the Morgan Hill corridor - PJ-0001, PJ-0002, PJ-0003 - represent poppy jasper, rather than some other gem material or a commodity category with no species behind it.

Standing: HYPOTHESIS, and it is the register's oldest unexamined assumption. Three entries have sat at moderate and speculative for the whole project on the strength of a dot.

Why it needs stating as a hypothesis at all. register/register.md's standing caution is blunt: I-909's entire gemstone legend reads "GEMSTONES / (dot) Locality." No species, no sub-type, no locality names, no table. All three symbols were examined at 4x from the 300-dpi scan and none carries a text label. The 12-page report names no gem species (controlled: jasp, chalced, agat, opal, chert all zero, against mercur = 28 present). So the cited source does not say what the gemstone is, and everything connecting these localities to jasper is inference from surrounding evidence.

What supports it - All three dots fall in the Morgan Hill corridor, and I-909 plots exactly three gemstone localities in all of Santa Clara County - all here. - I-909's unit list names "gemstones, jasper" as Franciscan commodities - as separate items, and the map's point symbology has no jasper category. - The material the corridor is known for is poppy jasper.

What contradicts it - The source names no species. That is not weak evidence for jasper; it is no evidence. - MRDS has a commodity code that would say orbicular jasper (GEM_SP) and the Morgan Hill records do not use it - an absence in the records, not in the ground, but worth weighing. - PJ-0003 sits in the Coyote basin on mapped landslide deposits, and now carries a dated 1950 non-find of jasper at Anderson Dam.

The decisive test is documentary, not field. USGS OFR 75-303 holds the CDMG Gemstones commodity summary that generated these dots - ILL 237598767, due before 2026-11-30. It would say what the commodity is. references/pending-documents.md calls it "the single highest-value document outstanding in the project", and this hypothesis is why.

Observation that would STRENGTHEN it. Orbicular jasper found at or near any of the three dot positions - though at +/-1000 m accuracy, "near" is doing heavy work.

Observation that would WEAKEN it. OFR 75-303 naming a different commodity; or an adequate Class B search at a dot position finding chert but no orbicular jasper.

Testability. - To strengthen: orbicular jasper at or near a dot position - gated - key: a position better than the +/-1000 m the dots carry, then access to whatever ground it names. All three are zone-extent and unresolved. - To weaken: OFR 75-303 naming a different commodity - in flight - ILL 237598767, due before 2026-11-30. The field alternative, an adequate Class B search at a dot, waits on the same position fix as the line above. - By document: OFR 75-303 - in flight - the same request. It settles this hypothesis outright and is the single highest-value document outstanding in the project.

Tested by: PJ-0003 primarily - it is the only one of the three with no other hypothesis attached, since H2 covers the two on fpv basalt and H7 was restated off it. PJ-0001 and PJ-0002 test this and H2.

Coverage check

Which register sites have a hypothesis attached, and which do not. D cannot be scored above 1 for a site with no test, so the gaps below are the remaining work.

All seventeen scorable entries now carry a test. The D gate is closed and scoring may begin.

PJ-0017 re-opened this gate when it was added on 2026-08-22 and H11 closed it the same day. Adding a register entry re-opens the D gate by construction - a new site with no test is a site whose D cannot be scored. Check this table after any addition.

One thing the coverage table makes visible. H3 and H6 have no register site - El Toro and the New Almaden nomenclature conflict are both settled by documents, not by walking. That is worth keeping in view: three of this project's twelve hypotheses cannot be advanced by any field trip, and treating them as fieldwork targets would waste days.

Testability at a glance

Can this project run each hypothesis's own tests, today? Statuses per GLOSSARY.md; the blocks above are authoritative, and tools/check-d-gate.py refuses a row that disagrees with its block.

What the table shows that twelve entries do not. Only H7 and H8 can be both strengthened and weakened by this project today. H4 can be confirmed from the public bridge and refuted only with permission - the shape that invites confirmation bias, named so it is not mistaken for progress. H1 has that shape in principle and for good: it can never be refuted, which is why it is kept as a constraint rather than worked as a question. Six hypotheses wait on the same kind of key - a private landowner in the Paradise Valley corridor - so the permission route is not one task among many; it is the critical path for half this file. And H10's first key is held by this project, which makes it the cheapest gate here.