The Destruction of 187J3X1 in the Spectroscopic and Photometric Record: A Comparison with Natural Stellar Death
Abstract
The destruction of the star 187J3X1 is the only dark forest strike that humans have observed and the empirical basis of the Dark Forest Deterrence, yet its spectra have not been examined in print and its photometry only for one statistic. We compare the archived monitoring of the star, 102 photometric epochs and 21 low-resolution spectra, with three families of natural stellar death drawn from 130 events in the Observatory's Crisis-Era transient catalogue. After the first departure the source rose for 17.4 ± 1.6 days to a peak 10.9 mag above the pre-strike star, declined at 1.38 ± 0.11 mag per 100 days after peak and at 1.05 ± 0.14 mag per 100 days at late times, and showed a nebular alpha-to-iron abundance ratio of +0.21 ± 0.14 dex. All four statistics lie inside the 95% predictive band of every family, and the amplitude lies inside the range of a luminous late-type progenitor. The chi-squared values on 4 degrees of freedom are 1.34, 0.87 and 4.33 (p = 0.86, 0.93 and 0.36), and the class weights run from 0.19 to 0.52. The nebular abundances and the light curve do not distinguish the event from a natural stellar death. The record could have flagged an abundance offset only above about 0.6 to 0.7 dex, the six pre-strike spectra are too sparse to fix the progenitor's class, and no signature of the unknown instrument was available to test. The evidence for the deterrence therefore rests on the prediction and its timing.
1. Introduction
The duty log of the Yunnan Astronomical Observatory for night 1,061 of its monitoring series on the star 187J3X1 holds a short entry in the hand of the astronomer on shift. The 59th exposure sequence had come back 0.31 mag brighter than the running mean, and the astronomer repeated it twice on the same night, before writing that the star was not as it had been. The 58 sequences before it, spread over nearly three years, showed a star that wandered by about two hundredths of a magnitude.
It has been more than half a century now since the destruction of 187J3X1 was observed, and the event remains the first and so far only observation of a dark forest strike, made on a third party. The star is the one whose coordinates Luo Ji broadcast in the spell (Wallfacer Program Archives, Records Office DE 33). The dark forest has not yet been tested by a strike on Trisolaris.
That single event is the empirical basis of the Dark Forest Deterrence (Strand-Okoye DE 56). The formal derivation of the cosmic-sociological argument for such a strike dates from the early Deterrence Era (Thorvaldsen DE 14). The literature has not examined the spectra, or set the light curve as a whole against natural stellar death, and the one reduction in print measured a decline rate alone (Kessendorp DE 49).
The matter examined here is whether the abundances and the light curve can tell the event from a natural stellar death. The instrument of the strike is unknown, so no one can say what it should leave in a spectrum, and the test we can run is one of exclusion: an event outside the known natural range of rise times, decline rates and ejecta compositions would have been distinguishable. Writing in Year 58 of the Deterrence Era (DE 58), we use only records and publications dated Year 57 or before.
Section 2 describes the archive, the sophon hazard and the statistics, and Section 3 sets out the measurements and the comparison; in Section 4 we state what follows for the deterrence, and in Section 5 we give the limits of the result.
2. Methods
The record is held in the Observatory archive as the monitoring series for 187J3X1 (Yunnan Astronomical Observatory DE 57). It comprises 58 photometric epochs before the first departure, taken between night 1 and night 1,043 at intervals of about 18 days, while the 44 epochs over the 372 days after it come at intervals that lengthened from one night to a mean of 8.7 days. They are 7 on the rise, 6 between the peak and the decline window, 17 in the decline window, 3 between that window and the late window, and 11 in the late window. Magnitudes are relative to the mean of the pre-strike epochs that scatter by 0.021 mag, a signal-to-noise of 52 per epoch. The photometer's limit, at a signal-to-noise of 5 in the sky-limited regime, where the noise does not change with the source, therefore lies 2.5 mag below the pre-strike star. The 21 spectra have a resolving power of about 2,000 and cover 3,900 to 6,200 Å. Six were taken before the departure at a signal-to-noise of 9 to 16 per resolution element (mean 12.8), and 15 after it, 4 of them early (day 9 to day 38) and 11 nebular (day 163 to day 341). Days are counted from t0, the first epoch more than 5 standard deviations (0.105 mag) above the pre-strike baseline.
We declare the sophon hazard to this record, as every archive of the period must. The whole of the series was recorded in the Crisis Era, before the Dark Forest Deterrence was established and before any window of partial suspension was announced, so there is no logged window to apply. A sophon can fix the logged result of any single measurement at a detector and sophons coordinate without measurable delay. The 102 photometric epochs and 21 spectra could therefore have been made to look like anything. We cannot test for that. We treat the archive as the Observatory delivered it and assume that sophon readership of it was continuous.
Four statistics enter the comparison, and the rise time is the interval from t0 to the peak of a smooth rise function fitted to the rise epochs (Castellanos-Yip DE 36), and the same peak gives the amplitude over the pre-strike mean, which is no comparison statistic because it depends on the luminosity of the progenitor. The decline rate is the least-squares slope of magnitude against time for the 17 epochs between 30 and 120 days after peak. The late rate is the same slope for the 11 epochs from day 170 to day 372, and the Student's t intervals of the two have 15 and 9 degrees of freedom. Nebular abundance is the mean over oxygen, magnesium, silicon and calcium of the abundance relative to the iron group, calibrated on nebular line fluxes (Linnekar DE 54). Its error is the 0.46 dex standard deviation between the 11 nebular spectra divided by the square root of 11 (t on 10 degrees of freedom). It allows for no offset common to the calibration, which the catalogue shares and which cancels to first order. Arguably the most exposed step is the zero of time, since an onset that preceded t0 by one cadence interval would lengthen the rise by up to 18 days, which in turn would move the rise time toward the slow end of every family. We therefore measured the comparators with the same 5σ threshold on light curves of similar cadence, and the ± 1.6 days below excludes that uncertainty.
Comparators are the 130 events of the Observatory's Crisis-Era transient catalogue (Yunnan Astronomical Observatory DE 52; it does not contain 187J3X1), with the same four statistics under the same definitions. They are 64 hydrogen-rich core-collapse events, 37 stripped-envelope core-collapse events and 29 thermonuclear events, classed by the catalogue from spectra at maximum light. Family means of peak absolute visual magnitude are -16.9 ± 1.1, -17.4 ± 1.2 and -19.2 ± 0.5. The catalogue's events were observed at higher signal-to-noise, with a median of 12 nebular spectra each. Their own abundance errors of about 0.06 dex are therefore small beside the family spreads of 0.16 to 0.21 dex. The 95% predictive band is the family mean plus or minus 1.96 times the square root of the family variance plus the squared error of our measurement.
The band holds the family standard deviation as known. For the thermonuclear family, a t on 28 degrees of freedom and the uncertainty of the family mean together would widen it by about 6 per cent, a change that alters no conclusion below. We treat the four statistics as independent within a family. A family's chi-squared is the sum of the four squared standardised offsets on 4 degrees of freedom. Its class weight is its likelihood divided by the sum over the three families, with equal prior weight. No family stands for the strike itself, since no one has predicted what it would leave in a spectrum, so the comparison tests exclusion from the natural range, beside the smallest offset detectable, taken as 2.80 predictive standard deviations (80% power at a two-sided 5% level), and every figure is computed from unrounded values.
3. Results
Before the departure the 58 photometric epochs scatter by 0.021 mag and show no trend (0.002 ± 0.003 mag per year). The 6 pre-strike spectra classify the star as late G to mid K. At a signal-to-noise of 9 to 16 they are too sparse and too noisy to measure the gravity-sensitive features that fix a luminosity class, and the class stays open (Abiodun-Ferrant DE 45).
A late G to mid K star can end in a natural hydrogen-rich collapse only as a supergiant, and a stripped-envelope or thermonuclear event needs a hot or degenerate progenitor, so those families fit only through a companion that we cannot exclude. We do not use this in the comparison, and the class weights below are weights on the light of the event, not on the star.
After t0 the source rose for 17.4 ± 1.6 days to a peak 10.9 ± 0.2 mag above the pre-strike mean. It declined at 1.38 ± 0.11 mag per 100 days between 30 and 120 days after peak (95% CI 1.14 to 1.62) and at 1.05 ± 0.14 at late times (0.73 to 1.37), and the nebular alpha-to-iron ratio is +0.21 ± 0.14 dex (-0.10 to +0.52). The 4 early spectra bracket the peak at a signal-to-noise of 8 to 14 and show a blue continuum with broad features. The catalogue classes by the hydrogen, helium and silicon lines at 6,563, 5,876 and 6,355 Å. The first and last lie beyond the 6,200 Å red limit of the spectrograph. The Balmer lines and the 5,876 Å helium line remain in range. At a signal-to-noise of 8 to 14 they cannot be separated from the broad features the spectra show, so they add no statistic.
Table 1 sets each statistic against the three families, and all four lie inside the 95% predictive band of every family. The largest standardised offset is -1.41, for the late decline against the thermonuclear family. The smallest p value for any single comparison is 0.157, uncorrected for the 12 comparisons, and after correction none approaches significance.
The combined chi-squared on 4 degrees of freedom is 1.34 for the hydrogen-rich family, 0.87 for the stripped-envelope family and 4.33 for the thermonuclear family (p = 0.86, 0.93 and 0.36). The class weights are 0.29, 0.52 and 0.19 (Table 2). The best-supported family outweighs the least by a factor of 2.7, which does not classify the event. With the catalogue's within-family correlations (none above 0.58 in magnitude) the chi-squared values become 1.32, 1.08 and 4.38 and the weights 0.31, 0.49 and 0.20, so independence does not drive the result.
Table 3 gives the union of the three 95% bands and the smallest offset detectable at 80% power. No statistic lies outside the union, and a departure in the abundance ratio would have been flagged only above 0.59 to 0.71 dex.
The amplitude bounds the progenitor and does not classify the event. Added to the catalogue's peak absolute magnitudes, it puts the star at -6.0 for the hydrogen-rich mean and -6.5 for the stripped-envelope mean, the luminosity of a supergiant. For progenitors of absolute magnitude -3 to -8 the three families give amplitudes of 6.7 to 17.2 mag at 95%, which contains 10.9, whereas a dwarf of +5 to +7 would give 21.9 to 23.9 for the hydrogen-rich mean. The run closed on day 372 with the source 6.6 mag above the pre-strike mean, after a fall of about 4.3 mag from the peak, so the calendar and not the photometer's limit ended the late window. The source stood 9.1 mag above that limit. At the late rate it would have reached it no earlier than about day 1,240, and later if the rate kept easing.
Kessendorp (DE 49) gave a decline rate of 1.55 ± 0.19 mag per 100 days over the interval from the peak to 100 days after it. Refitted over that interval, which holds the 6 transition epochs and 13 of the 17 decline-window epochs, the archive gives 1.49 ± 0.10. The window therefore accounts for 0.11 of the 0.17 difference from our value of 1.38, and the remaining 0.06 lies well inside the earlier error. The two fits share 13 epochs, and we do not treat the estimates as independent.
4. Discussion
The nebular abundances and the light curve do not distinguish the destruction of 187J3X1 from a natural stellar death. Each of the four statistics falls inside the natural range of every family. The amplitude falls inside the range for a luminous late-type progenitor, and the weights are too close to rank the families. The hypothesis that the event can be told apart on these data is not supported.
We write within the Luo-Ji / Ye framework, where an exposed position invites a strike whether or not anyone lives there and no spectral signature of the strike is required. That makes the broadcast and the wait the working evidence, since they fix the star as a target before the event and the event as an outcome after it, which in turn leaves the spectra adding nothing to the argument and taking nothing from it.
For the deterrence the consequence is plain. The event is one observation, made on a third party, and these statistics of its light cannot show that it was a strike and not a death. What carries the evidential weight is the prediction and its timing, which are the spell and the wait. We offer no rescue of the spectral case.
A record of a strike on Trisolaris would meet the same obstacle, since the instrument is unknown and no signature has been predicted for it. Until one is stated, the light of a destroyed star cannot carry the argument.
5. Limitations
Sophons set the first limit: a sophon can fix the logged result of any single measurement, and nothing in the archive can show whether it did, so a record made to resemble a natural death would give the result we found, and so would a record of one.
Progenitor class is the largest limit. The pre-strike archive holds 6 spectra at a signal-to-noise of 9 to 16 per resolution element, or 32 if all are co-added. That is too sparse and too noisy to fix the spectral type and luminosity class of the star, and the amplitude of 10.9 mag fits a supergiant but cannot establish one. A secure classification of the star as a low-mass dwarf would exclude every natural channel at once. No star of that mass ends in a collapse or a thermonuclear event on its own. The observed 10.9 mag is about 11 mag short of the 21.9 mag that a natural peak would show above a dwarf. That is the next measurement, and it needs a deeper pre-strike spectrum from an archive that we do not hold.
Comparators and photometry set the rest, since the 130 catalogue events were selected by having been detected and classified in the Crisis Era, few have a measured progenitor and so give no distribution of amplitude about a known star, and the 29 thermonuclear events estimate the within-family correlations only weakly. The class weights depend on equal prior weight and Gaussian families, and they measure how little the record separates the families. The onset is uncertain by up to one pre-strike cadence of 18 days, which the rise time inherits.
What a strike by an undescribed instrument leaves in a star's light is still not completely understood. On the present record the data favour only a statement of what they cannot do, and the entry for night 1,061 is the entry that any observatory would have made for a natural transient. The next set of observations will add to that entry only if a signature is stated before the event it is to test.
References
- Thorvaldsen, E. (14 DE). Suspicion between civilisations that cannot verify one another, a derivation from the two axioms. Journal of Cosmic Sociology, 2, 31-64.
- Wallfacer Program Archives, Records Office (33 DE). Action and observation logs of the four Wallfacer files, with evidence registers. Wallfacer Program Archives, declassified series, Files 1–4, action-log series.
- Castellanos-Yip, R. (36 DE). Rise times and decline rates of transients measured from threshold-crossing epochs in sparse monitoring series. Proceedings of the Reconstituted Chinese Academy of Sciences, 72, 150-171.
- Abiodun-Ferrant, P. (45 DE). Spectral classification of late-type stars from sparse, low signal-to-noise spectra, and the limits of luminosity-class assignment. Proceedings of the Reconstituted Chinese Academy of Sciences, 74, 120-141.
- Kessendorp, A. (49 DE). Photometric reanalysis of the first Yunnan monitoring run on 187J3X1. Yunnan Observatory Historical Papers, 12, 31-52.
- Yunnan Astronomical Observatory, Stellar Spectroscopy Group (52 DE). Crisis-Era transient catalogue, light-curve and nebular-spectrum statistics for 130 classified events. Yunnan Astronomical Observatory Archive, catalogue CET, release 1.
- Linnekar, D. (54 DE). Alpha-to-iron-group abundance ratios from nebular line fluxes at low resolving power. Proceedings of the Reconstituted Chinese Academy of Sciences, 75, 301-326.
- Strand-Okoye, H. (56 DE). The single observed strike as the empirical footing of the deterrent. Deterrence Era Strategic Studies, 16, 5-30.
- Yunnan Astronomical Observatory, Stellar Spectroscopy Group (57 DE). Monitoring series for 187J3X1, photometry and spectra of the Crisis-Era run, Deterrence-Era reductions. Yunnan Astronomical Observatory Archive, series 187J3X1, release 2.
Open in Uncited Press →