An argument is as strong as its weakest load-bearing assumption, and an honest argument lays those assumptions on the table — with the mark that says how firm each one is, and with the falsifier that would break it. This register gathers every assumption on which the Wednesday of 25 April AD 31 rests, in the order of the book's two axes, and it ends with the measure that applies on the other side. Nothing here is new; it stands scattered through the chapters. Setting it together is the test: whoever finds the weakest row and breaks it, breaks the case.
D.0 · How to read this register
Every assumption carries one of the three marks of this book. CALCULATED — recomputable; you reproduce it yourself. DOCUMENTED — seen in the primary source, with location. CONSTRUED — a reading or a choice; it weighs as much as you grant it, and no grain more. The final column is the falsifier: the concrete find that would break the row. An assumption without a falsifier is not a foundation but an article of faith, and those stand apart here (D.3).
D.1 · The day-axis — that the crucifixion fell on a Wednesday
assumption
bears
mark
what would break it
"Three days and three nights" (Matthew 12:40) is a measure (three full days), not a loose figure of speech
ch. 3 — the entire day-axis
CONSTRUED (reading), supported by the D-corpus [CALCULATED]
one clear pre-modern use of "three days and three nights" — in the corpus of the idiom itself, not in a later exegetical harmonisation about it (appendix B §B.3c weighs the candidates put forward) — for demonstrably less than three full twenty-four-hour days
The double construction never counts inclusively anywhere in the corpus searched
ch. 3, appendix B
CALCULATED (base rate)
a single attested inclusive use of the double construction within the searched corpus itself — a later interpreter who argues it inclusively (Augustine, B §B.3c) does not count here, and that is expressly conceded
This book cites the received text (Stephanus 1550 / Scrivener 1894), collated against eight printed editions
Not breakable but payable, and the price is stated with it: in Luke 24:21 the received text reads σήμερον more, and that makes the heaviest counter-verse of this book sharper (ch. 13). And it is not everywhere the majority text: in Acts 12:25 it stands against the Byzantine majority. Whoever chooses a text-form keeps it also where it goes against him
The chronology of Paul (Acts and Galatians) counts nowhere in this book — neither as relief for AD 31 nor as an objection to AD 33
ch. 16, seventh open flank
CONSTRUED (self-binding)
a dated, independent fixed point in Acts 1–9 that pins the conversion before or after AD 31. This row is an obligation, not an assumption: whoever declares an instrument blunt against his own year may not reload it against another's year
The day of death is the fourteenth of Nisan (erev Pesach), not the fifteenth
ch. 5
CONSTRUED, strongly supported [DOCUMENTED] (b. Sanh. 43a; the working-day actions)
compelling proof that the last meal was the official temple seder of 15 Nisan
The passion week carried two sabbaths (festival sabbath + weekly sabbath) with a working day between them
ch. 4
CONSTRUED (reconstruction)
proof that the spice sequence (Mark 16:1 / Luke 23:56) only adds up under a single sabbath
The weekday of every date is exact
ch. 7, appendix A
CALCULATED
an error in the Julian day count (checkable for yourself)
The seven-day week ran unbroken from the first century to today, and the planetary day names of the Roman world coincide with it, so that a Julian date gives its weekday by continuous count
ch. 7, appendix A
CONSTRUED, supported [DOCUMENTED]: Vettius Valens' worked example names Mechir 13 of Hadrian's fourth year, 8 February 120, as the day of Hermes, a Wednesday on the continuous count (Pingree I.9, pp. 25–26 = Kroll I.10, p. 26; Neugebauer & Van Hoesen, Greek Horoscopes, L 120,II, p. 116). ⚠ One datum stands against it: the Pompeii graffito CIL IV 4182 names 6 February 60 dies Solis, which is a Wednesday on the continuous count. An explanation from the naming of the day by its first hour exists (Brind'Amour), but it is called possible and speculative in the literature (Bultrighini & Stern, 2022); the graffito is booked here as an open datum, not explained away
a dated first-century document whose weekday name departs from the continuous count and cannot be explained from the naming of the day — the graffito comes close
The first-century temple observed (visibility), did not compute (molad) — the weekday follows the visible crescent, not the computed moon
ch. 7, ch. 8, ch. 14
CONSTRUED, strongly supported [DOCUMENTED] (the molad calendar is demonstrably of later date). ⚠ And this is why it remains an assumption: the mishnaic description was set down around 200 and its court is that of Yavneh, not that of the temple; the step back to the first century is an assumption of continuity
proof that the first-century court computed (molad) rather than observed — under that reading AD 30 (14 Nisan = Wed 5 April) becomes a Wednesday and AD 31 loses it (Mon 26 March). ### ⭐ Since 2025 this falsifier has a published advocate, and that belongs here. André Reis appeals to Sacha Stern, to the Elephantine papyri, to Segal (1957) and Sprengling (1911) to argue that the calendar could already have been calculated in the first century, and he places 1 Nisan of AD 31 on a predicted new moon rather than on an observed crescent. He has not succeeded, but he is the first to try it seriously, and the register would be dishonest if it pretended no one tries. ⛔ The objection, however, is that his own principal witness says the opposite. Stern writes on p. 103 that the rabbinic calendar rested "originally" on the appearance of the new moon and "later" was replaced by a calculated calendar, and announces in the next paragraph that early Jewish lunar calendars rested "in general" on the new crescent; the calculated scheme of "others" that he adduces on p. 159 he weighs on p. 160 as "would hardly have been functional" and "a marginal opinion with which the majority of rabbis disagreed", nowhere in the Palestinian Talmud. And on pp. 162–163 he draws his own conclusion: Judaea in the Hasmonaean, Herodian and early Roman periods had "a lunar, empirical calendar". ⚠ The same page, however, also bounds this book: whether the procedures of m. Rosh ha-Shanah and the Tosefta were ever fully carried out Stern calls "anyone's guess". He gives the nature of the calendar, not its execution — and that is exactly why this row remains an assumption and does not become a finding. [DOCUMENTED — Stern (2001), pp. 103, 159–163, read at first hand; Reis (2025), pp. 14–15]
The sheaf falls on 16 Nisan (the Pharisaic count), not on the Sunday after the weekly sabbath
ch. 4, ch. 8
CONSTRUED (a choice within an old dispute) ⚠ and the self-diagnosis belongs with it: on this book's own chronology 16 Nisan is a Friday, and only the Sunday count (the Sadducean) brings the firstfruits to a Sunday. Whoever chooses that count because it comes out right does what ch. 1 and ch. 15 hold against the consensus
proof that the first-century temple followed the Sadducean count, or that 23:11 can only be the weekly sabbath
The day did not begin everywhere alike: in Galilee at morning, in Judea at evening
ch. 5 — the link ch. 5 itself calls "the most composite"
CONSTRUED (a reconstruction, not a measurement)
no demonstrable double day-beginning in first-century Judea and Galilee — then the distinction falls, and with it the harmonisation of the supper
The top two rows together bear the day-axis; the bottom nine connect it to the text, the week and the observational calendar. The most vulnerable is the first — the reading of the sign — and precisely for that reason the base rate that hardens it stands beneath it. To the row of the unbroken week belongs one fringe theory named by name, because it is sometimes offered as an ally of the Wednesday: the "lunar sabbath", a sabbath that shifts with the phase of the moon. This book does not adopt it; it undermines the fixed weekly sabbath on which the day-axis rests (ch. 4), and an ally that breaks the row it should lean on is no ally.
D.2 · The year-axis — that that Wednesday fell in AD 31
assumption
bears
mark
what would break it
AD 31 had an intercalary month (Adar II), so that Nisan became the April moon. Acknowledged and fixed here: without it the March moon stays and 14 Nisan falls on Tuesday 27 March — not on a Friday. In AD 31 the choice of rule thus decides which midweek day it becomes, not whether there is a Friday (ch. 8, ch. 15)
ch. 8
CONSTRUED (not documented in a primary source)
a source ruling out a Jewish Adar II in AD 31. More precise than it stood before, and the reason is worth mentioning for honesty's sake: Parker & Dubberstein show no Babylonian intercalary month in AD 31 (S.E. 342) — the April placement runs there via an Ululu II in the autumn of AD 30 (appendix E.2). That is documented and it burdens this row, but it does not break it, for the Babylonian intercalation is not the Jewish one: the council in Jerusalem inserted only an Adar. Whoever wants to break this row must therefore touch the Jewish month
The calendar followed the visible crescent after the equinox, not the conjunction or a fixed rule that excludes April
ch. 8, appendix C
CONSTRUED (a choice among documented rules). ⚠ A second practice is documented: in Egypt, from Aristobulus to Philo, the Passover fell at or around the equinox, on average a month before the Babylonian date (Stern, 2000, p. 171 n. 69); that Jerusalem kept the late, Babylonian practice rests on few Judean data and is an assumption (ch. 8)
proof that first-century Judea bindingly followed a single rule that excludes the April moon
The first crescent of AD 31 appeared on the evening of 11 April
appendix C
CALCULATED (Yallop q = +0.32, zone A)
a visibility model that places the crescent of 11 April below the threshold
The baptism fell in the autumn of 27
ch. 9
CONSTRUED (mode of counting) — two convergent but individually reversible lines: Luke 3:1 on the autumn count (the weakest of the three counts; the best-attested gives 28/29) and John 2:20, a reversible anchor
compelling proof that Luke's fifteenth year demands the sole-reign count. Registered more precisely than it stood, and it costs: the autumn-inclusive count is attested for Tiberius by no document at all, and it is carried as unattested, not refuted: not probable, but not impossible — no surviving witness numbers his regnal years inclusively (Steinmann, 2022a, p. 115), the one Jewish numbering that can be read (Ant. 18.106) is non-inclusive, and the two coin corpora that might have tested it cannot, because at both new years the accession stub had elapsed before the news of Augustus' death could arrive (ch. 9). The one document registered in advance as decisive, P. Mur. 18, came back undecided: the regnal formula is there, but day and month both lie in the lacuna (DJD II, p. 103). ⚠ The same count carries the surviving AD 30 configuration; whoever breaks it breaks both (ch. 14). The count has two frames, the Syro-Macedonian and a Jewish Tishri year (ch. 9). A Jewish rule for Tishri exists in the Bavli (b. RH 3a, 8a), against a Nisan rule in the Yerushalmi (y. RH 1:1), both amoraic; a document dated by Tiberius in either frame does not. ⚠ And the choice is a convergence choice: the count is taken because John 2:20 agrees with it, and the alternative of a baptism in 29 is set aside on a ground that is circular with respect to the ministry length derived from it (ch. 9)
The ministry counted four Passovers (~3½ years), not three
ch. 9, ch. 14
CONSTRUED — carried as support and not as compulsion, on the early reception [DOCUMENTED] (Irenaeus; Eusebius, Dem. Ev. 8.2) and on a test registered in advance whose outcome was possible, not compelled (appendix E.3)
proof that John 5:1 is not a Passover and that no hidden year sits in the chain. The frozen test has been run and the result stands here: of the three criteria for the extra spring, 1 and 2 were not met and 3 was met only in its own wording, as support; on this axis AD 31 is therefore admissible and not indicated — and AD 30, which needs exactly three Passovers, no more so (appendix E.3)
The bottom two rows are the hinge between AD 30 and AD 31: a baptism in 27 with four Passovers yields AD 31, with three Passovers AD 30 (ch. 14). Both rows have been weakened since this register was first drawn, and both weakenings are entered above rather than in the chapters alone. The top two are the narrowest link of the entire book — the calendar.
D.3 · The presupposition that is not an assumption under the clock
One thing bearing the case is no measurement and has no falsifier, and it belongs on its own, out loud.
That God delivers to the day — that a sign comes out exactly. This is not a calendar assumption but a hermeneutical presupposition (ch. 1, ch. 2). It cannot be broken by a find, for it is not an empirical claim; it is the lens through which the book reads the sign as a measure. The book does not hide this: under this presupposition the Wednesday wins decisively; without it, it falls back into a three-way race (ch. 16, appendix F). Whoever does not share the presupposition may leave the measuring core (calendar, astronomy, chronology) standing untouched — that core does not lean on it. [CONSTRUED — and marked as such, not sold as proof.]
D.4 · The two sharp breaking points, and the measure on the other side
Of all the rows above, two are sharp and concrete — not hermeneutics, but a find that would settle the case at a stroke:
1. a source ruling out a Jewish Adar II in AD 31 (D.2, row 1 — see there why the Babylonian series does burden this row but does not break it) — then the Wednesday falls, and with it the whole date; 2. one clear pre-modern use of "three days and three nights" — in the corpus of the idiom itself, not in a later exegetical harmonisation about it (appendix B §B.3c weighs the candidates put forward) — for demonstrably less than three full twenty-four-hour days (D.1, row 1) — then the sign falls, and the field returns to the Friday.
Both lie open on the table; both are testable. That the case names its own breaking points is not its weakness but the very condition that makes it testable.
A third lies close by: the unbroken week (D.1), against which one dated graffito already stands open (CIL IV 4182).
And the measure applies on both sides (ch. 8, ch. 15). Every demand this register makes of AD 31, it makes in the same breath of the rivals:
- AD 33 requires that that year had no intercalary month and that the March moon became Nisan despite a 1 Nisan before the equinox — two calendar choices the consensus rarely accounts for. - AD 30 requires exactly three Passovers (John 5:1 not a Passover) and a weekday that depends on the visibility criterion (Thursday or Friday) — a choice, not a measurement.
Whoever holds these rows against AD 31 but lets them pass for AD 30 and AD 33 measures with two measures; and then the certainty of the consensus is a majority, not a measurement.
D.5 · What this register does not deliver
- It does not weigh the rows. It says which assumptions bear weight and how firm each is, not how heavily they weigh against one another; that weighing stands in Chapter 16 and appendix F. - "CONSTRUED" is not a disqualification. A construed reading may be strongly supported (the fourteenth of Nisan), convergent but reversible (the baptism in 27), possible but not compelled (the four Passovers), or thin (the undocumented Adar II). The mark states the kind of certainty, not the degree; that stands in the falsifier column and in the chapters. - It does not register the findings, only the assumptions. What a search actually returned — against the case and, twice, in favour of it — stands in appendix E, with the deposited register and its year-axis entries (appendix E.5, E.6). - The list is the load-bearing core, not every detail. The corroborating rings — the rabbinic front (ch. 5), the patristics (ch. 6), the earth as witness (ch. 12) — do not bear the date and do not stand here; they appear in the weighing only under "corroboration, each with its own base rate" (ch. 16).
An argument is as strong as its weakest load-bearing assumption, and an honest argument lays those assumptions on the table — with the mark that says how firm each one is, and with the falsifier that would break it. This register gathers every assumption on which the Wednesday of 25 April AD 31 rests, in the order of the book's two axes, and it ends with the measure that applies on the other side. Nothing here is new; it stands scattered through the chapters. Setting it together is the test: whoever finds the weakest row and breaks it, breaks the case.
D.0 · How to read this register
Every assumption carries one of the three marks of this book. CALCULATED — recomputable; you reproduce it yourself. DOCUMENTED — seen in the primary source, with location. CONSTRUED — a reading or a choice; it weighs as much as you grant it, and no grain more. The final column is the falsifier: the concrete find that would break the row. An assumption without a falsifier is not a foundation but an article of faith, and those stand apart here (D.3).
D.1 · The day-axis — that the crucifixion fell on a Wednesday
The top two rows together bear the day-axis; the bottom nine connect it to the text, the week and the observational calendar. The most vulnerable is the first — the reading of the sign — and precisely for that reason the base rate that hardens it stands beneath it. To the row of the unbroken week belongs one fringe theory named by name, because it is sometimes offered as an ally of the Wednesday: the "lunar sabbath", a sabbath that shifts with the phase of the moon. This book does not adopt it; it undermines the fixed weekly sabbath on which the day-axis rests (ch. 4), and an ally that breaks the row it should lean on is no ally.
D.2 · The year-axis — that that Wednesday fell in AD 31
The bottom two rows are the hinge between AD 30 and AD 31: a baptism in 27 with four Passovers yields AD 31, with three Passovers AD 30 (ch. 14). Both rows have been weakened since this register was first drawn, and both weakenings are entered above rather than in the chapters alone. The top two are the narrowest link of the entire book — the calendar.
D.3 · The presupposition that is not an assumption under the clock
One thing bearing the case is no measurement and has no falsifier, and it belongs on its own, out loud.
D.4 · The two sharp breaking points, and the measure on the other side
Of all the rows above, two are sharp and concrete — not hermeneutics, but a find that would settle the case at a stroke:
1. a source ruling out a Jewish Adar II in AD 31 (D.2, row 1 — see there why the Babylonian series does burden this row but does not break it) — then the Wednesday falls, and with it the whole date; 2. one clear pre-modern use of "three days and three nights" — in the corpus of the idiom itself, not in a later exegetical harmonisation about it (appendix B §B.3c weighs the candidates put forward) — for demonstrably less than three full twenty-four-hour days (D.1, row 1) — then the sign falls, and the field returns to the Friday.
Both lie open on the table; both are testable. That the case names its own breaking points is not its weakness but the very condition that makes it testable.
A third lies close by: the unbroken week (D.1), against which one dated graffito already stands open (CIL IV 4182).
And the measure applies on both sides (ch. 8, ch. 15). Every demand this register makes of AD 31, it makes in the same breath of the rivals:
- AD 33 requires that that year had no intercalary month and that the March moon became Nisan despite a 1 Nisan before the equinox — two calendar choices the consensus rarely accounts for. - AD 30 requires exactly three Passovers (John 5:1 not a Passover) and a weekday that depends on the visibility criterion (Thursday or Friday) — a choice, not a measurement.
Whoever holds these rows against AD 31 but lets them pass for AD 30 and AD 33 measures with two measures; and then the certainty of the consensus is a majority, not a measurement.
D.5 · What this register does not deliver
- It does not weigh the rows. It says which assumptions bear weight and how firm each is, not how heavily they weigh against one another; that weighing stands in Chapter 16 and appendix F. - "CONSTRUED" is not a disqualification. A construed reading may be strongly supported (the fourteenth of Nisan), convergent but reversible (the baptism in 27), possible but not compelled (the four Passovers), or thin (the undocumented Adar II). The mark states the kind of certainty, not the degree; that stands in the falsifier column and in the chapters. - It does not register the findings, only the assumptions. What a search actually returned — against the case and, twice, in favour of it — stands in appendix E, with the deposited register and its year-axis entries (appendix E.5, E.6). - The list is the load-bearing core, not every detail. The corroborating rings — the rabbinic front (ch. 5), the patristics (ch. 6), the earth as witness (ch. 12) — do not bear the date and do not stand here; they appear in the weighing only under "corroboration, each with its own base rate" (ch. 16).