Tiberius, John, and the years of the ministry

The third count — Syrian-inclusive, fifteenth year = autumn 27–autumn 28 — is the manner of counting by which this book extracts the baptism of 27 from Luke 3:1, and honesty demands that I show its foundation, including where it is brittle. It appears at its sharpest in the scholarly literature in Hoehner (1977), who calls it "method 3" and dates the fifteenth year to "21 September 27 to 8 October 28". Walk through his footnotes, and they do not carry that date. His load-bearing note invokes Mommsen (Römisches Staatsrecht II.2, 802–804) for the claim that Syria counted the Roman emperors from the autumn — but those pages in Mommsen deal with the Bildnissrecht, the right to the imperial portrait on the coinage; there is no Syria there, no Seleucids, no new year, no regnal-year count. Cichorius (1923), Hoehner's own authority for the exact dates, leans in the same question on precisely the same Mommsen pages — two footnotes, independently, to a passage that does not carry it. And Kraeling (1942), the best-founded defender of "27/28", concedes in his own note 18: there are no coins proving that Tiberius was counted that way; the double dating that pins the autumn boundary is Nero's, not Tiberius'.

This I must count in full against my own side, for I lay the same measure on both. The only direct, double-dated Tiberius piece — the Antiochene coin (RPC I 4270), "(year) 1 [...] (Actian year) 45" — dates Tiberius' year 1 to AD 14/15, and thereby sets the fifteenth year, on its own Actian calendar, at 28/29, not at 27/28. The inclusive 27/28 reading, which would carry AD 31 on a three-and-a-half-year ministry, is thus the reading the only double-dated piece does not give. How much that coin can say is limited, and the dossier "the reckoning flank" below sets out why: it measures the pagan provincial count, and there the news of Augustus' death arrived after the new year, so that inclusive and non-inclusive reckoning give the same year 1; it fixes the era, it does not discriminate between the systems. Unattested, then, and not refuted: not probable, but not impossible; and on the fifteenth year itself the best-measured reading (28/29) stands on the side of AD 33, not of this book. I do not choose 27/28, therefore, because it rescues AD 31 — that would break the standing rule. What the find does yield is modest and symmetrical: the heaviest chronological entry against the Wednesday, Luke's fifteenth year, is even in the standard monograph no clean measurement but a footnote chain that ends in coin-portrait law. The fifteenth year pins nothing down, and the decision between the years does not fall here, but on the day (see the main text above). [DOCUMENTED — Mommsen II.2³ 802–804 and Kraeling, ATR 24 (1942) 346, seen at first hand; RPC I 4270 at first hand; the conversion CALCULATED]

The ground both invoke is an old tradition, and this book does not lean on it. Blinzler and Finegan both appeal to Luke as a Syrian of Antioch — a reading that exists, that has not been refuted, and that makes a Syro-Macedonian autumn count more natural in his hands. It is named here and not carried. Its witnesses are late and their dating is disputed, and an origin is not a mode of counting. This book's AD 27 baptism rests on the autumn count itself — booked as unattested and not refuted, in the workbench on the reckoning flank below — and not on Luke's birthplace. Whoever accepts the Antiochene origin gains nothing here that the book needs; whoever rejects it costs the book nothing. [CONSTRUED]

The starting point of the temple. Herod's eighteenth year = 20/19 BC (Josephus, Ant. 15.380). What anchors it is not a counting convention but Josephus' own sequence: Ant. 15.354 puts Augustus' arrival in Syria in the seventeenth year, that visit is fixed on the Roman side at 20 BC (Dio 54.7), and the eighteenth year follows directly. Add forty-six years (there is no year zero) = AD 27/28 → first Passover in the spring of 28.

The rival figure. Josephus gives two. Bellum Judaicum (BJ) 1.401 puts the rebuilding in the fifteenth year. Counted from the appointment (40 BC), the 18th year is 23/22 and the 15th 26/25; counted from the capture of Jerusalem (37 BC), the 18th year is 20/19 and the 15th 23/22. Only the 18th year on the capture meets the Syrian visit. And the fifteenth-year figure serves nobody's reckoning: add 46 and you land in AD 21/22 or AD 24/25 — too early for every candidate, for AD 30 as much as for 31 and 33. Whoever lets it count is left with no John 2:20 arithmetic at all.

What follows. The 46-year route is less sharp than every party treats it — this book included, up to this line. It stays here as the second anchor beside the autumn count, but without the word "fixed": the starting point hangs on Josephus' sequence in Ant. 15.354, not on the one figure this book happened to pick. [DOCUMENTED — Ant. 15.354, 15.380 and BJ 1.401 checked in the primary text; CONSTRUED — the weighing]

The addition towards the crucifixion year. The year depends on two choices — the baptism and the number of Passovers:

The lower bound. Pilate took up office by 26/27 at the latest (Josephus, Ant. 18.32–35); a baptism in the autumn of 27 falls neatly behind that. The book's chronology is thus: baptism autumn 27, first Passover 28, four Passovers, crucifixion AD 31 — verifiable, but dependent on two construed choices, and that is stated honestly beside it. [CALCULATED]

The reading that inserts a spring between John 2:13 and John 6:4 carries, on this book's route, the whole difference between AD 30 and AD 31. It was therefore put to a test whose criteria were written down before the sources were read: three that would have to be met for the extra spring, three that would tell against it. No criterion was adjusted along the way. The outcome is possible, not compelled, and it is entered here in that form.

Of the three for, criterion 1 was not met — a constraint that would force the plucking of the grain (Matt. 12:1; Mark 2:23; Luke 6:1) after the Passover of John 2:13 independently of the assumed Judaea–Galilee order. The synoptists themselves disagree on where the incident belongs: Matthew sets it after the Baptist's question from prison, Mark and Luke long before, ahead of the calling of the Twelve. At least one of them is not ordering chronologically here, and Mark 2:1–3:6 is a thematically arranged series of controversies. Criterion 2 was not met either — a seasonal reading of John 4:35 ("yet four months, and the harvest") that would exclude the proverb reading. The seasonal reading turned out stronger than expected, and no such proverb is attested anywhere; but the commentary tradition is divided and does not divide in a way that settles it. Criterion 3 was met — the early reception of John 5:1 — and in its own wording, as support. Of the three against, two were met (the Judaea–Galilee order rests on a harmonisation the synoptists do not claim; the third leg re-uses the material of the first) and one was not: the tradition is not predominantly proverb-reading, and it does contain readers who take the four months as a real interval and build no chronology on it.

And the author on whom the reading leans claims no more than that. Hoehner writes that the plucking of the grain "would fit well" between the two Passovers (1977, p. 56). That is not a compelling argument, and he does not offer it as one. Further: his two supports cannot both be used. He reads John 5:1 "with great hesitation" as Tabernacles rather than as a Passover, calls Blinzler's claim that a three-year ministry requires 5:1 to be a Passover "false", and his Tabernacles argument rests in turn on the seasonal reading of 4:35 — "If 5:1 were the next Passover, it would come immediately after 4:35" (1977, pp. 58–59). Whoever runs the Hoehner package cannot run Irenaeus at the same time, for Irenaeus' whole point is that 5:1 is a Passover. This book therefore carries one route and not both: the early reception, as support, with Irenaeus' own untenable conclusion named beside it — and not Hoehner's ordering as an independent leg.

What that costs, said plainly: the four-Passover reading may not be presented as compelled, and earlier versions of this book presented it more strongly than the test bears. What it does not cost: the reading is not special pleading. It is a respectable reading that merely does not force. And the result runs the same way in both directions — on this axis AD 31 is admissible and not indicated, and AD 30, which needs exactly three Passovers, is equally not indicated. The axis is undecided, and the decision falls elsewhere. [DOCUMENTED — Hoehner, 1977, pp. 56 and 58–59, seen at first hand; the criteria fixed before the reading]

The reversal reading above hangs on one lexical hinge: that "temple" in John 2:20 is the Greek naos, and that naos strictly means the sanctuary building — completed as early as around 18 BC, so that "forty-six years" would point to a moment in the past rather than to a building still in progress. The distinction naos (the sanctuary) versus hieron (the whole complex) is a real tendency in the New Testament. But whether it compels here can be tested against the witness who stands closest to it: Josephus, who describes precisely this temple in Greek. And his own usage does not carry the strict reading.

Josephus does narrow naos in the technical building account — "the naos itself was built by the priests in a year and six months" (Ant. 15.421). But that very sentence undermines the reversal reading: were "forty-six years" the building time of the naos itself, it could not be right, for that sanctuary stood in a year and a half. And elsewhere Josephus uses the word broadly: "they build the naos [...] and have drawn the colonnades and the walls around it" (Ant. 11.87) — a naos with colonnades and walls is the whole temple work, not the cella alone. The lexica point the same way: the foundational dictionary (LSJ) gives as the first sense of naos "the temple" (the building as a whole); the narrow sense "cella" is the derived, second sense, and the New Testament restriction is a contextual particularity, not a property of the word.

This tips "forty-six years" back to the majority reading — the temple the speakers had watched rising before their eyes for forty-six years (and longer: the work continued until about AD 63) — and with it to the first Passover in 27/28 and the crucifixion in 30/31. It makes AD 33 harder, but — and this is the honest limit — it does not exclude it: the grammatical supports of the reversal reading (the dative as a point in time, the aorist as completed) stand apart from the lexis and survive. John 2:20 thus remains what the note on the sources says it is: a reversible anchor, which under the majority reading fits AD 31 and can be bent towards AD 33 by a practised opponent — no hard anchor, but one that, measured against Josephus' own usage, leans slightly towards the pair 30/31. [DOCUMENTED for Josephus' usage — Ant. 15.421 and 11.87 (Greek); CONSTRUED for the weighing]

Reis (2025, p. 3) reckons along the same axis, but from the other end. He places the start of the building at the end of 21 or the spring of 20 BC, has the work finished some 18 months later (18/17 BC), and counts the forty-six years of John 2:20 from that completion: the first Passover then falls in AD 30 — the opening of a ministry that, in his chronology, ends on the cross in 33. That start — "ten years after Actium" — is not Josephus' figure but Reis' own inference: Josephus dates the beginning by a regnal year alone, and against his own anchor (Ant. 15.354) Reis comes out one year early; his result does not change for it. His AD 30 is a first Passover, then, not the crucifixion year that chapter 14 weighs. And those 18 months are Josephus' figure for the naos itself (Ant. 15.421) — the very datum this workbench measures above. His variant therefore adds no new evidence; it is the reversal reading counted in the other direction. This book counts the forty-six years from the start of the work and arrives at 27/28; he counts them from the end of the same year and a half and arrives three years later. Laid beside Steinmann's reckoning, his is not a second witness for the reversal reading but the same one: both rest on the narrow naos, and that narrow sense is, measured against Josephus' own usage, the derived meaning and not the first. [DOCUMENTED for Reis' reckoning — Reis (2025), p. 3; CONSTRUED for the weighing]

Behind the sentence "Egyptian inscriptions and papyri, finding year marks running exclusively from 14/15 to 36/37" lies a documentary corpus worth writing out, for it is the count nearest to Luke's own world. In the Egyptian papyrus and ostraca material, collected in Bureth's standard repertory, Tiberius' first year stands at 14/15 — and not from an editorial decision alone: under "year 1 of Tiberius" stand some fifteen dated pieces. Had Egypt applied the ordinary "stub" convention (the accession year as year 1, however short), that year 1 would cover only the ten days between Augustus' death (19 August 14) and the Egyptian new year (1 Thoth, 29 August) — and fifteen documents from ten days, in a province that could not yet have received the news from Rome, is not credible.

That the stub convention was the norm in Egypt — Nero's fourteenth year, the year of his death, is abundantly attested there — makes the exception under Tiberius precisely explicable, and one papyrus supplies the mechanism. P.Oxy. VII 1021 announces the death of Claudius in Oxyrhynchus as news and is already dated "year 1 of Nero" — thirty-five days after the death, and still as news. Applied to Augustus' death, the news reaches Egypt only in late September 14 — after 1 Thoth. There was no stub year to open, for by the time it was known, the new Egyptian year had already begun. No anomaly, but arithmetic.

That same count stands on coin in Judaea itself. The prefects Valerius Gratus and Pontius Pilate struck at Jerusalem on Tiberius' regnal years: Gratus' years 2 to 11 (AD 15–24), Pilate's years 16 to 18 (AD 29–31) — and each of these falls within their terms of office known from Josephus only if Tiberius' year 1 falls in 14/15. Under a count from the co-regency (12/13), Gratus' "year 2" would fall in AD 13, when he was not yet prefect and Tiberius not yet emperor. The direct, primary count thus sets Tiberius' epoch unanimously at 14/15 — what the Excursus established numismatically now also stands documentarily (and, as set out there, does not carry the co-regency as the counter of the AD 27 baptism). [DOCUMENTED — Bureth (1964) 25; P.Oxy. VII 1021 and RPC I 4958–4969, seen at first hand; the conversions CALCULATED]

Beneath the epoch of Tiberius — and beneath the dating of Herod's death — runs a second front, and it is textual. Josephus writes that Philip died "in the twentieth year of Tiberius' rule", after "thirty-seven years" of tetrarchy (Ant. 18.106). Philip's own coins bear that 37th year (RPC I 4952–4953, on the era beginning at Herod's death in 4 BC), and so the passage anchors that year of death — the anchor that carries ch. 10. Whoever wants to move Herod to 1 BC therefore needs a different reading, and it exists: part of the Latin tradition reads "the twenty-second year" (Filmer, 1966; Beyer, 1998). Were it to stand, the anchor would fall — and so it too belongs tested here, with the same rigour I lay against an opponent elsewhere.

It does not stand, and Niese's critical apparatus says why. "εἰκοστῷ" — the twentieth — is uniform in all Greek witnesses (the manuscripts A, M, W and the Epitome); the only witness for "twenty-two" is the old Latin translation. More than that: Niese's asterisk before the lemma is his own sign that the Epitome — a Greek branch independent of the main text — reads with the text at this point, and it thus positively confirms "twenty" and "thirty-seven". And the Latin branch that supplies "twenty-two" supplies in the same sentence not "thirty-seven" but "thirty-two" tetrarchy years (other Latin manuscripts: "thirty-five"). No one — not a single witness — reads "22 + 37". And precisely that pair of numbers is what the early dating needs; it is a contamination, the year from the Latin and the regnal figure from the Greek. A witness that demonstrably garbles the second number of the sentence does not carry the first. The "twentieth" stands, and with it the Philip anchor of ch. 10 — even though the epoch of 14/15 that follows from it does not lie to this book's advantage. And the passage costs more than the epoch. Twenty plus thirty-seven from 4 BC gives 33/34, which on an inclusive count would be the twenty-first year: the sentence is a numbered regnal year of a Roman emperor, written by a Jewish narrator, reckoned non-inclusively. It is the sharpest single witness against the autumn-inclusive count on which this book's baptism of 27 depends, and it is weighed as such in the dossier on the reckoning flank below rather than left standing here as an anchor only. [DOCUMENTED — Niese, Opera IV (1890) 160 with apparatus, seen at first hand; Jachowski, 2015]

This is the heaviest open entry in the book, and it is set down here in the only form it can be carried.

What is measured. Every attested reckoning of Tiberius' regnal years puts year 1 in AD 14/15 — Antioch (RPC I 4270), Seleucia (RPC I 4330), Alexandria (RPC I 5075–5105, the series running to year 23 and stopping there). None puts it in 13/14. *(The literary witnesses — Philo, Josephus, Tacitus, Dio — are of a different kind and are not counted in that list: they give the reign's duration, twenty-two and a half to twenty-three years, and a duration says how long a reign lasted, not how a dating formula numbers its fifteenth year. They fix the epoch at 14, and the Kenner box above cites them for exactly that; they do not fix the numbering. The distinction is entered against my own side elsewhere; it is entered here as well, because it holds in both directions.)* The autumn-inclusive count from which this book takes the baptism of 27 is therefore attested nowhere for Tiberius. Doig, the one author who reaches autumn 27 along it, has an analogy with Herodian practice and no document, and concedes that the Roman coins reckon dynastically. And the one modern defender of an inclusive count from 14, Meier, lands on AD 28 and not on autumn 27 — established here on secondary reports of his position and not on his own pages, and marked as such.

Why the word is nevertheless unattested and not refuted. What the autumn count needs is that Tiberius' accession stub was itself numbered year 1. Neither corpus can test that. Augustus died on 19 August 14; the Egyptian new year (1 Thoth) fell on 29 August and the Actian new year on 2 September, and the news cannot have reached Alexandria or Antioch before either — P.Oxy. VII 1021, in the dossier above, still reported such a death as news thirty-five days later. Both cities therefore begin numbering at the first new year after the news arrived, and that yields year 1 = 14/15 under inclusive and non-inclusive reckoning alike. The coins fix the era; they do not discriminate between the systems. What they do exclude is a stub numbered retroactively, before the news had come; the autumn count does not need that. The only calendars on which the question lived — the Dios autumn, new year 20 October, and a Jewish Tishri year, 1 Tishri on 14 October on the Babylonian months, both after the news had come — have left no document dated by Tiberius at all. So the corpus is structurally unable to answer the question that is asked of it, and absence of evidence cuts both ways here, exactly as this book insists elsewhere that it does. [DOCUMENTED — RPC I 4270, 4330, 5075–5105; P.Oxy. VII 1021; CALCULATED — the news interval and the new-year arithmetic]

What stands on the other side is Jewish, and the fullest survey does not contain it. Steinmann concludes that "there is no evidence for inclusive reckoning of Tiberius's reign" (2022a, p. 115). That conclusion holds for the corpus he searched, and that corpus is Roman, Greek and Egyptian: the Latin historians, Philo, Josephus, Alexandrian and Syrian coins, Latin and Greek inscriptions. The Jewish halakhic material does not appear in it — and that material prescribes precisely the reckoning he finds no trace of, for precisely the same use: the dating of a document. m. Rosh ha-Shanah 1:1 sets the new year for kings on 1 Nisan; the Gemara asks what that serves halakhically, and Rav Ḥisda answers li-shetarot, "for documents". Then the baraita: "a king who acceded on the twenty-ninth of Adar — once the first of Nisan arrives, a year is reckoned to him; and if he acceded only on the first of Nisan, no year is reckoned to him until the next Nisan comes" (b. Rosh ha-Shanah 2a). One day of reign before the new year counts as a whole regnal year. That is non-accession, inclusive reckoning, written out, with its application attached.

Two links stay open, and I name them before an opponent finds them. First, the date of the source: the Mishnah is around 200, and the baraita, though presented as tannaitic, stands in the Bavli; that the rule already held in the first century is plausible and not demonstrated. Second, the application to a foreign ruler: the rule is framed for kings of Israel, and whether Jewish datings of a Roman emperor followed it is a separate question. Neither gap is filled by wanting it filled.

And a third point follows from the rule itself, and it cuts against me. m. Rosh ha-Shanah 1:1 counts from 1 Nisan. Applied to Tiberius, who acceded on 19 August 14, it makes the stub up to Nisan 15 his first year and puts his fifteenth year from Nisan 28 to Nisan 29 — not at the autumn of 27. The rule cited here for the principle of inclusive reckoning does not by itself give this book's date.

On the second link the rabbis speak, and they disagree. The Bavli limits the Nisan rule to Israel's own kings: "lo shanu ella le-malkhei Yisra'el, aval le-malkhei ummot ha-olam mi-Tishri maninan", "Rav Ḥisda said: they taught this only for the kings of Israel; but for the kings of the nations of the world we count from Tishri" (b. Rosh ha-Shanah 3a, derived from Nehemiah 1:1 and 2:1, Kislev and Nisan in the same twentieth year of Artaxerxes). At 8a Rav Pappa reads the Mishnah's "1 Tishri is the new year for years" in the same way, as the rule for documents, and the Gemara reconciles the two: Nisan for the kings of Israel, Tishri for the kings of the nations. On that rule a Roman emperor's fifteenth year runs from Tishri 27, and the news-rule of the coin corpora does not change that: on the Babylonian months 1 Tishri AD 14 fell on 14 October (Parker & Dubberstein), after the news of 19 August can have reached Judea. The Yerushalmi says the opposite: "af le-malkhei ummot ha-olam ein monin ella mi-Nisan", "R. Eleazar in the name of R. Ḥanina: for the kings of the nations of the world, too, one counts only from Nisan" (y. Rosh ha-Shanah 1:1), derived from Haggai 1:15 and Zechariah 1:1, the sixth and the eighth month in the same second year of Darius. Three limits belong with that. Both rules are amoraic, of the third and fourth centuries, and both are derived from the dates of Persian kings in Scripture, not from a practice under Rome. The Bavli must itself keep its rule standing against Haggai and Ezra with ad hoc answers — Cyrus counted like an Israelite king because he was virtuous, Cyrus, Darius and Artaxerxes one man, Nisan before his fall and Tishri after it (3b–4a) — and at 8a two other amoraim read "for years" without documents at all. And the Palestinian voice, the nearer one to Luke's Judea, is the one that says Nisan. Stern's reading of P. Mur. 18 below, Nero counted from Nisan, stands on that side. The Jewish material thus offers both a Tishri 27 and a Nisan 28, it does not choose between them, and a Tishri count for a Roman emperor is available on it, not attested. [DOCUMENTED — m. RH 1:1; b. RH 2a, 3a–4a, 8a; y. RH 1:1 (Sefaria); CALCULATED — the year count]

On that second link there is published expert support — and a published expert against it, in the same breath. Sacha Stern, discussing P. Mur. 18, holds that "the second regnal year of Nero may have begun, according to the Jewish computation, in Nisan of 55 CE", and notes: "Nero became emperor on 13 October 54 CE, but this would not normally have been reckoned as the starting-date of his regnal years (pace Benoit, Milik, and de Vaux 1961: 103) [...] For rabbinic evidence of regnal years beginning in Nisan, see M. RH 1:1, Y. RH 1:1 (56b), B. RH 2a–b" (Calendar and Community, pp. 92–93 with n. 333). Nero acceded on 13 October 54; if his second year began in Nisan 55, his first was the stub from October to Nisan. That is inclusive reckoning of a Roman emperor's years in a Jewish document, held by the standard monograph on the Jewish calendar. Against it stands the editor of that very document: "'An deux de Néron' correspond à 13 octobre 55 – 12 octobre 56" (Benoit, Milik & de Vaux, DJD II, p. 103) — the accession-anniversary count, from the man who had the papyrus in his hands. And Stern writes "may have begun", and adds himself that "the possibility of scribal error must always be taken into account". Support, then, and not establishment. And support for the principle only: Stern's Nisan count gives the fifteenth year from Nisan 28, not from the autumn of 27 (above).

The test that was registered in advance came back undecidable, and that is entered here in full. Before the text was opened I set down what would settle it: Mur 18 should carry a regnal formula, and its month would separate the two readings, since a date in the spring or summer of 55 is year 2 only on Stern's reckoning and still year 1 on Milik's. The formula is there. The month is not. Milik's own translation reads "[Le [...] du mois [...] a]n deux de Néron César" — the square brackets are his restoration, and day and month both lie in the lacuna (DJD II, p. 103, seen at first hand). The document that was to decide cannot decide — not "not yet", but not at all. I record the registered test as undecided, in the same place and the same type in which I would have recorded it had it fallen my way.

And the sharpest thing against my own route is Jewish too, and it stands a paragraph above. Ant. 18.106 dates Philip's death to the twentieth year of Tiberius, after thirty-seven years of tetrarchy from 4 BC — that is 33/34. On an inclusive count 33/34 would be the twenty-first year. Here, then, a Jewish narrator gives a numbered regnal year of a Roman emperor in a synchronism, and he numbers it non-inclusively. It is the nearest thing in this whole dossier to the case Mur 18 was to have supplied, and it runs against me. The halakhic rule shows that the reckoning my chain needs was written down somewhere; Ant. 18.106 shows that the one Jewish writer whose numbering can actually be read did not use it. Both belong on the page, and the second is the reason the flank stays open. The balance, then: the autumn-inclusive count is not probable, but not impossible. [DOCUMENTED — Steinmann, 2022a, p. 115; m. RH 1:1 and b. RH 2a in the Hebrew and Aramaic; Stern, pp. 92–93 with n. 333; DJD II, p. 103; Josephus, Ant. 18.106 with Niese's apparatus]

One word on the author I lean on here, for he excludes my year by name. Steinmann's conclusion is that the evidence "calls into question AD 30 or 31 (or any year before AD 30) as possible years of Jesus's crucifixion", and that "the testimony of the combined evidence from the Gospels for a ministry of about three and a half years supports AD 33 as the correct year of the crucifixion" (2022a, p. 116). I use his measurement and not his outcome, and the reason that is permitted can be checked by anyone with the article open. The two arguments rest on different material and neither derives from the other: the first is a count of coins, papyri and inscriptions; the second is a claim about the length of the ministry, and it enters his own concluding sentence as a premise — "since the Gospels' testimony [...] indicates a ministry for Jesus of over two years" — not as a result of the survey. Whoever accepts the first is not thereby bound to the second. That the second tells against me is registered here and in the dossier on the extra spring, in the main text and not in a footnote.