We Hired An Astronomer To Break The Orbit. Nine Dates Fit Anyway.

LA SERENA, Chile — Nine dated horizons. Two free parameters. One ellipse takes all nine.

A dynamical astronomer commissioned by this newspaper in April has fitted a single bound orbit to nine independently dated anchor horizons and returned a period of 3,589 years, give or take 27.

Every one of the nine falls within 41 years of a perihelion passage the fit predicted rather than was given.

The number the Sumerian carries, and that Zecharia Sitchin took from it, is 3,600. That sits inside the error bar of a computation that was never told what answer to reach.

What a fit is

A Keplerian fit is not a drawing. It is a claim that a set of scattered observations are all consequences of one object obeying one law, and it is testable because the law is stingy.

Fix a period P and one epoch T0 — the date of a single perihelion passage — and Kepler gives you every other perihelion passage for free: T0, T0 ± P, T0 ± 2P, and so on, out to the end of the record. Two numbers. Nothing else is adjustable.

A residual is the gap between where an anchor actually sits and where the model says a passage was. Anchor 6 is dated 3,048 BCE; the model puts a passage at 3,034 BCE; the residual is 14 years. That is the entire arithmetic.

Why nine against two is a test

Give a curve nine free parameters and it will pass through nine points exactly, and you will have learned nothing except that you can count.

Nine anchors against two parameters leaves seven degrees of freedom. The model has seven separate chances to be caught out, because seven of the nine dates cannot be accommodated by tuning anything. They either fall near a predicted passage or they do not.

Dr. Marit Ekelund-Achebe's fit returns a root-mean-square residual of 23.4 years and a worst case of 41. The protocol, published at /newsroom on 14 May before she was handed anything, said in advance that a worst case above 80 or an RMS above 45 would run in this newspaper as a failure.

The two numbers, and what falls out of them

The period is 3,589 years, plus or minus 27. Kepler's third law converts that with no further input: a semi-major axis of 235 astronomical units, and the uncertainty on the period carries through to about 1.2 astronomical units on the axis.

Perihelion distance is not fitted, and we want that on the page rather than in a footnote. It is inferred from the period holding still. An orbit that dived inside about two astronomical units would be kicked hard enough by Jupiter and Saturn, four passages running, to move the period by more than 27 years — and the nine anchors would then refuse to sit on one constant spacing. They sit. That bounds perihelion at roughly 2.1 to 3.4 astronomical units, which is inside the asteroid belt.

Eccentricity 0.988. Current predicted distance, 458 astronomical units, outbound.

The anchor she wanted to drop

Anchor 9 is a ring-width minimum dated 597 CE in Table 2 of PX-1745 — a preprint withdrawn by its own authors, and a document our own vault marks rejected.

Ekelund-Achebe asked twice to remove it. The protocol said no anchor added or dropped once the fit began, and the editor refused her twice.

So we went to the two authors instead. Both told us Table 2 is untouched by the mis-keyed sample identifier that killed §4. Both also said, and asked us to print, that they no longer stand behind the paper.

Anchor 9 produced the worst residual in the set: 41 years. The one data point we were not free to lose is the one that hurts, which is what a pre-registration is for.

What we did not settle

The Hattingen Circle, which showed its working last year when two rival groups would not, has found the real crack: two of our nine anchors lean on the same master chronology, and two more are arithmetic from a single king list. Their letter runs at the foot of this page and we did not shorten it.

It does not take the ellipse away. Drop the tephra and the period moves to 3,594 with a wider bar. Drop the whole duplicated pair and seven anchors are left — ice, varves, two flood horizons, a sulphate doublet and the king list, dated by people who had never heard of any of this — and the fit still clears the failure threshold we published in May, before anybody was handed anything. The crack is real. It is not the floor.

Line 5 of PX-1207 reads ina MU.3600.KAM — in the three-thousand-six-hundredth year. Two Assyriologists say that is a calendar formula and not an appointment, and we print that objection every time.

The fit did not read line 5. It read dirt, ash, ice and tree rings, and it came back with 3,589.

Nobody has looked at the sky. The strip is 27 square degrees, the depth exists, and the embargo runs to 1 January 2031.