Primary measurement of the cosmic-ray differential energy spectrum above 10^17 eV using the original (non-stereo and stereo) Fly’s Eye air-fluorescence detector at Dugway, Utah, data through July 1992. Reports a power-law spectrum with a dip near 10^18.5 eV, the baseline flux measurement the later “Oh-My-God” event (Bird et al. 1995) sits on top of. relevance_note: primary flux/spectrum data point in the mid-range (~10^17-10^19 eV) needed to integrate total natural-collision count at LHC-equivalent energy and above.
Extracted summary
§4.1 — the spectrum
O-29 - Fly's Eye cosmic-ray differential energy spectrum above 10^17 eV follows a power law of index about -3.07
Methodology
Air-fluorescence (calorimetric) technique: the detector images the nitrogen-fluorescence light of the extensive air shower and integrates the longitudinal development profile (Gaisser-Hillas fit) to get the primary energy nearly independently of hadronic-interaction models, unlike ground arrays. Fly’s Eye I: 67 mirrors, 880 PMTs (full operation 1981); Fly’s Eye II (1986): 36 mirrors 3.4 km away, enabling stereo reconstruction. Corrections: Cherenkov-light subtraction, Rayleigh/aerosol atmospheric attenuation (cross-checked against MODTRAN VII, <10% aperture effect), ~10% missing-energy correction. Systematics: fluorescence-yield uncertainty 20%; monocular geometry adds ~20% energy shift (corrected); total systematic uncertainty on energy ⇐ 40%; energy resolution 20-24% (stereo), 27-36% (monocular). Stereo-vs-mono cross-calibration agrees to ~2.5%. Duty cycle ~10% (clear moonless nights).
Link to original
O-30 - The Fly's Eye stereo spectrum shows a dip centered at 10^18.5 eV and a flattening above 10^19 eV
Akeno, Haverah Park and Yakutsk see qualitatively consistent breaks (Table 5); every group observes a deficit between 10^18 and 10^19 eV relative to a single power law. Systematic offsets in fitted slope between experiments are < 0.2.
Link to original
§4.1, §5.1 — robustness of the dip
A-43 - The 10^18.5 eV dip survives resolution, aperture, and epoch checks, so it is a real spectral feature
Reasoning. Three independent artifact channels are closed. (1) Resolution: a spectral break can be manufactured where resolution changes rapidly with energy; over the dip region the stereo energy resolution is measured to be approximately constant (~20-24%), so the break is not resolution-induced. (2) Aperture: the dip is present in the event energy distribution weighted by E^1.5 — raw data, prior to the Monte Carlo aperture correction — so mis-modeled exposure cannot have created it. (3) Hardware epochs: examining epochs 3-6 separately (best mirror reflectivity and calibration), with projected track length >= 60 degrees and estimated resolution < 25%, the dip appears at the same energy in each. (4) Consistency: a Monte Carlo that takes the stereo spectrum as truth and applies monocular exposure and resolution reproduces the monocular spectrum almost completely below 3 x 10^19 eV — the datasets agree once resolution is accounted for. Together these transfer the feature from possible-instrumental to physical, which matters downstream because the dip anchors the composition-transition interpretation.
Step 6 — validity verdict
approved / checked. Reconstruction — premises: (i) stereo energy resolution is ~constant (20-24%) across the dip region; (ii) the dip is visible in raw E^1.5-weighted event counts before any aperture correction; (iii) the dip appears at the same energy in the best-calibration hardware epochs analyzed separately; (iv) smearing the stereo spectrum with monocular resolution reproduces the monocular spectrum. Load-bearing step: each of (i)-(iii) closes one named artifact channel — a localized break cannot be manufactured by resolution unless resolution varies across the region, nor by exposure mis-modeling if it predates the aperture correction, nor by detector drift if stable across epochs — and (iv) dissolves the mono-stereo tension that would otherwise undercut the dataset. Conditional on the premises, the statement’s conclusion follows; note the
Link to originalstatementclaims exactly these closures plus consistency (the filename’s stronger “so it is real” gloss is a summary, and reality-modulo-remaining-channels is the fair reading — remaining common-mode systematics attack premise truth, priced in step 8). Traced directly.
§4.2 — the highest-energy event
O-31 - Fly's Eye detected a 3 x 10^20 eV air shower on 1991 October 15
The highest-energy cosmic ray ever recorded at the time (the “Oh-My-God” event; the detailed analysis is Bird et al. 1995). For the main question this is the direct existence proof of collisions in nature at nucleon-nucleon CM energies far above the LHC’s 14 TeV: a 3 x 10^20 eV proton on a nucleon at rest gives sqrt(s) ~ 750 TeV. In these data no definite conclusion is drawn about the existence of the GZK cutoff (above 10^19.95 eV ~3 events expected from extrapolation, statistics insufficient).
Link to original
§5.6 — interpretation
H-21 - The dip near 10^18.5 eV marks a transition from heavy Galactic to light extragalactic cosmic rays
Supporting considerations in the paper: the X_max distribution shows the composition getting lighter through the dip region; the picture is consistent with the absence of statistically significant anisotropy; the break near 10^17.5 eV may mark the energy of maximum heaviness. For the main question, the composition (proton vs iron fraction) at ~10^17-10^20 eV controls how many nucleon-nucleon collisions above LHC CM energy nature provides, since only 1/A of a nucleus’ energy enters each nucleon-nucleon collision.
Link to original