Very high energy gamma rays from the direction of Sagittarius A*

Bottom line: this is the measurement that settled the TeV Galactic Centre, and it holds up. H.E.S.S. sees a 6.1σ and a 9.2σ excess centred within about 30″ of Sgr A*, consistent with a point source (σ_source < 3′ at 95% CL), with a hard power law (α = 2.21 ± 0.09 ± 0.15) and 1.8×10⁻⁷ m⁻²s⁻¹ above 165 GeV. The CANGAROO-II steep spectrum (α ≈ 4.6) does not survive this; the paper's remark that H.E.S.S. would have seen their low-energy flux in minutes is correct and worth taking at face value. The Whipple point is handled honestly too — the authors note the factor-3 discrepancy and decline to claim variability nobody has independently seen. What is done well: two data sets with different thresholds analysed separately and shown to agree; Crab and PKS 2155-304 used as reconstruction checks; muon calibration; statistical and systematic errors quoted separately; and an explicit statement that coincidence with Sgr A* is not proof of origin. The dark-matter section is framed as a constraint conditional on the fitted cutoff, not a discovery claim, which is the right posture. Soft spots, in order of size. First, the ring background. The Fig. 1 caption admits a Galactic-Plane feature whose significance is 'under investigation', and §2 never states the ring radii. A ring averages over azimuth, which helps, but in a field with a bright diffuse ridge the leakage term is not obviously small, and the 25% systematic is attributed mainly to energy calibration with no background term. That touches the flux normalization and, less so, the index — which is what feeds the CANGAROO comparison and the proton-luminosity estimate. It does not threaten the detection. Second, the source-identification ambiguity is acknowledged and correctly left open; later H.E.S.S. work resolved it into a composite, so nothing here is wrong, just incomplete. Third, the 4 TeV cutoff lower limit is fit-dependent and the Mχ > 12 TeV bound inherits that; the paper says as much. Who should read it: anyone working on the Galactic Centre, on TeV instrumentation, or on indirect dark matter. It is a benchmark measurement paper. Recommendation: send it out. Referees should ask for the ring geometry and a background systematic estimate; everything else is in order.

The paper's central claim has two parts: a significant point-like excess at Sgr A* and a power-law spectrum with index 2.21 and flux 1.82e-7 above 165 GeV. The detection (6.1σ and 9.2σ) is robust to background misestimation because the excess is strong. The load-bearing vulnerability is the flux normalization and spectral index, which are derived by subtracting a ring background in a field where the paper itself reports a feature extending along the Galactic Plane whose significance is 'under investigation' (Fig. 1 caption). The ring radius is not stated in §2, and the paper does not quantify how much diffuse Galactic-Plane emission leaks into the ring. If the ring overestimates the local background, the flux is underestimated; if it underestimates it, the flux is inflated and the spectrum is artificially hardened (a flat background component pulls the fitted index down). The quoted 25% systematic is attributed mainly to energy calibration and does not include a background-model term. This matters because the flux normalization is what drives the CANGAROO comparison and the physical interpretations (proton luminosity, dark-matter limits) in §3. The authors acknowledge the Galactic-Plane feature as unresolved, so the concern is not that they hid it; it is that the quantitative impact on the flux and index is left open.

The reader's weakest_assumption — the ring background in a non-uniform diffuse field — is the right target. The paper's detection claim is well supported: two independent data sets, 6.1σ and 9.2σ, a position consistent with Sgr A* within errors, a point-source angular distribution, and cross-checks on the Crab Nebula and PKS 2155-304. The instrument calibration is verified with muon images and cosmic-ray rates. The spectral index of 2.21 ± 0.09 ± 0.15 is far from CANGAROO's 4.6 ± 0.5, so the qualitative conclusion (hard spectrum, inconsistent with CANGAROO) is robust to a 25-50% flux error. The vulnerability is quantitative: the flux normalization and the exact index depend on the background model, and the paper does not quantify the diffuse Galactic-Plane contribution to the ring. The authors flag the Galactic-Plane feature as 'under investigation' (Fig. 1 caption), so they acknowledge the limitation; what remains unresolved is its numerical impact. Because the detection and the qualitative spectral conclusion survive, the verdict stays ACCEPT; the concern is a qualification on the flux normalization, not a threat to the central claim. I would not downgrade to CONDITIONAL unless the concrete test shows a shift larger than the quoted systematic.

If the measurement holds, this is the first solid TeV-band detection of the Galactic Centre region and the first measurement of its spectrum at these energies, and it is a result of obvious and lasting importance for the field. The analysis is well documented for its era: two data sets with different thresholds and different trigger conditions give consistent spectral indices (2.11 +/- 0.19 and 2.21 +/- 0.09), which is a meaningful internal cross-check; the instrument is calibrated with muon images and with cosmic-ray rates; and the energy reconstruction and flux determination are validated end-to-end on the Crab Nebula, where the reconstructed index (2.63 +/- 0.04) and flux agree with published values. Statistical and systematic errors are quoted separately throughout, the source-extension limit is stated as a falsifiable number, and the interpretive sections are clearly separated from the measurement. The exclusion of the CANGAROO-II spectrum, if correct, is a substantial result in its own right, and the paper states it plainly while remaining courteous about the discrepancy.

The H.E.S.S. collaboration set out to establish whether the Galactic Centre emits very high energy gamma rays at all, and if so with what spectrum, using two imaging atmospheric Cherenkov telescopes in Namibia during the summer of 2003. They report a point-like excess coincident with Sgr A* to within about half an arcminute, significant at $6.1$ $\sigma$ in the first data set and $9.2$ $\sigma$ in the second, with a power-law spectrum of index $2.21$ and a flux above 165 GeV of about 5% of the Crab Nebula's. The point of the paper is as much the shape as the detection: the spectrum is far harder than the steep $E^{-4.6}$ spectrum reported by CANGAROO-II from 2001-2002 data, and the authors argue that the low-energy flux implied by that result would have appeared in H.E.S.S. within minutes of observing. If the measurement holds, the Galactic Centre joins the small set of established TeV sources, the emitting region is compact (under 3 arcminutes, under 7 pc at the distance of the centre), and the spectrum shows no cutoff below 4 TeV and no sign of variability over the campaign.

What carries the argument. The instrument is the stereoscopic imaging atmospheric Cherenkov technique: two telescopes record the same shower, the images are parameterized by their centres of gravity and second moments, and the geometry is reconstructed in three dimensions, giving roughly $0.1^{o}$ angular resolution per gamma ray and 15-20% energy resolution from image intensity. Gamma-ray candidates are selected by image shape, with loose cuts rejecting 96% of cosmic-ray background while keeping 50% of gamma rays. The background is estimated from a ring around the assumed source position, and the signal is the excess over that ring. Calibration rests on muon images for absolute photon detection efficiency and on cosmic-ray rates for agreement with simulations; the whole energy reconstruction chain is validated by recovering the Crab Nebula's known index of $2.63$ and flux above 1 TeV of $(1.98 +/- 0.08) x 10^{-7} m^{-2} s^{-1}$, and the pointing by reconstructing the positions of the Crab and PKS 2155-304.

Core claim. Using the stereoscopic reconstruction of shower images from two H.E.S.S. telescopes, the authors measure a localized excess of gamma-ray candidates centred at RA $17^{h}45^{m}41.3^{s}$, Dec $-29^{o}0'22''$, offset from Sgr A* by $12''$ in Galactic longitude and $14''$ in latitude, both within the quoted $30''$ uncertainty. The combined data give a power-law spectrum $F(E) = F_0 E^{-2.21}$ with $F_0 = (2.50 +/- 0.21) x 10^{-8} m^{-2} s^{-1} TeV^{-1}$ for the July/August set and a consistent index of $2.11$ for the June/July set, a flux above the 165 GeV threshold of $(1.82 +/- 0.22) x 10^{-7} m^{-2} s^{-1}$, and a lower limit of 4 TeV on any exponential cutoff. The angular distribution of the excess is consistent with a point source, giving an upper limit of $3'$ (95% CL) on the Gaussian source size. The authors set this measurement directly against the CANGAROO-II sub-TeV spectrum and the Whipple flux above 2.8 TeV, noting that the three results cannot all describe a steady source.

The premise it rests on. The background is estimated from a ring around Sgr A* in a field where the authors themselves report a feature extending along the Galactic Plane whose significance is still under investigation, so any diffuse emission leaking into that ring would shift the quoted flux and significance.

The Galactic Centre becomes a confirmed TeV gamma-ray source, so any model of Sgr A* or its surroundings must now supply particles reaching multi-TeV energies at the observed rate.

The hard index near 2.2 is close to what shock acceleration produces, which favours a shock origin over the steep cutoff expected from some accretion-flow models.

The compactness limit of under 7 pc, combined with a hard spectrum at a source age of order $10^4$ years, implies particle diffusion in the central region is much slower than in the Galactic disk.

Supersymmetric dark matter annihilation as the dominant source is disfavoured, since the 4 TeV cutoff limit pushes the neutralino mass above 12 TeV.

The CANGAROO-II spectrum is excluded for a steady source, so either that measurement is wrong or the source changed by a large factor between 2002 and 2003.

The paper's ring background sits inside a field where the authors themselves report an unquantified excess extending along the Galactic Plane, so the flux normalization is the number most exposed to a better diffuse-emission model, even if the detection itself survives.

A source this compact and this hard invites a search for counterparts at other wavelengths within the few-arcsecond error box; the paper overlays a Chandra map but does not claim a multi-wavelength identification.

If the emission is hadronic and tied to the young supernova remnant Sgr A East, the required proton energy of order $5 x 10^{47}$ erg is modest enough that similar remnants elsewhere in the Galaxy should be detectable TeV sources, which is a checkable prediction for the same instrument.

The absence of variability in a single season does not exclude the year-scale changes the CANGAROO and Whipple comparison hints at; a multi-year light curve from the same telescope would settle whether the discrepancy is instrumental or physical.

This review was written by a machine. No person has reviewed this paper on Pith yet. A scientist listed on Pith can review it for a fee, and their review is published here whatever it concludes.

Request a human review