{"id":"0858be95-c286-4443-9c3f-0cb7ec5bcd8d","arxiv_id":"2501.00787","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Measurements from 57 AU show a cosmic UV background of about 221 photon units at 1000 Å and 264 at 1500 Å after accounting for dust, about half of which is unexplained.","lead":"New Horizons' Alice spectrograph, observing from 57 AU, measured the cosmic ultraviolet background at high galactic latitudes in two bands and found an unexplained glow after subtracting known sources. The result includes the first firm measurement of the background between 912 and 1100 Å, where prior missions only had upper limits.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 912–1100 Å first-detection claim rests on Box data whose 172 Å FWHM resolution is comparable to the band width; without LSF deconvolution, red-wing leakage from longer wavelengths can mimic a 1000 Å offset.","rationale":"The reader's weakest assumption was the GALEX single-scale-factor cross-calibration, which is a valid systematic concern for the absolute zero-point and would inflate the quoted uncertainties. However, an even more fundamental issue is that the Box's 172 Å FWHM is nearly as wide as the 912–1100 Å band itself, so the '1000 Å' offset is a convolution of the true CUVB over a broad wavelength range. Because the 1100–1350 Å region is blanked and the 1400–1800 Å offset is 264 photon units, red-wing leakage can plausibly contribute tens of photon units to the 912–1100 Å measurement. The paper's own caveat that the spectral decline from 1100 to 912 Å may be an artifact of the Box line width is precisely the expected signature of such contamination. The high-resolution Stem channel provides a clean spectral measurement but with an uncertainty of ±48 photon units, too large to independently confirm the Box's 221 ± 11 precision. A forward-model deconvolution using the measured LSF can settle whether the 912–1100 Å detection survives; until that test is performed, the paper should remain conditional. If the deconvolved offset remains near 200 photon units, the first detection stands; if it collapses toward zero, the result is an instrumental artifact. This does not change the reader's conditional verdict but identifies a distinct, more direct threat to the central claim.","tokens_in":23020,"tokens_out":21986,"duration_ms":207031,"concrete_test":"Perform a maximum-likelihood deconvolution of the Box spectra using the measured 172 Å FWHM line-spread function (obtained from the LyACAL 2007/2023 template or stellar calibration observations), modeling the zero-reddening offset as a free function of wavelength and explicitly including leakage from the 1100–1800 Å continuum. If the inferred 912–1100 Å offset is consistent with zero (or below ~150 photon units) at 2σ, the claimed 221 ± 11 offset is not a firm first detection of that band.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The most load-bearing weakness is spectral isolation, not absolute calibration. The Box has a FWHM of 172 Å (§2.1), nearly the full 188 Å width of the claimed 912–1100 Å band, and the analysis never deconvolves the line-spread function; §4.4 notes the detector region was integrated down to 850 Å to capture the short-wavelength signal. Counts in the 912–1100 Å pixels therefore include red-wing leakage from the 1100–1350 Å region (blanked in Fig. 15) and potentially from 1400–1800 Å, where the offset is 264 photon units. The Lyα template of §3.3 removes only scattered Lyα, not the astrophysical continuum at those longer wavelengths. The paper itself states that the decrease in offsets from 1100 to 912 Å 'may be an artifact of the Box line width'—the signature of red-wing contamination. The high-resolution Stem gives 172 ± 48 photon units, far less precise than the Box's 221 ± 11, so the first-firm-detection claim is carried by a measurement whose spectral purity is not demonstrated; the quoted ±11 excludes LSF modeling error.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper presents new observations of the cosmic ultraviolet background (CUVB) with the Alice spectrograph on New Horizons at ~57 AU from the Sun, targeting high-latitude fields with |b| > 40°. The authors subtract detector dark counts (measured with the door closed and interleaved with sky exposures) and a Lyα scattering template built by differencing 2007 and 2023 blank-sky observations. They then correlate Alice surface brightness in two bands (912–1100 Å and 1400–1700 Å) with Planck E(B−V), finding zero-reddening offsets of 221 ± 11 photon units at 1000 Å and 264 ± 24 photon units at 1500 Å. The former is claimed as the first firm detection of the offset in the 912–1100 Å range. Comparing with known contributors (unresolved galaxies, stars, O VI/C IV line emission, two-photon emission) leaves an unexplained excess of about 133 and 103 photon units in the two bands. An upper limit of 3.2 ± 3.0 photon units is reported for 600–800 Å emission below the Lyman limit.","tokens_in":23243,"tokens_out":5236,"duration_ms":50142,"significance":"If correct, this is a landmark measurement: it would be the first firm detection of the diffuse ultraviolet background in the 912–1100 Å band from outside the solar system, and the agreement with earlier GALEX-based offsets at 1500 Å would validate the method. The paper is careful in its dark subtraction, uses interleaved dark exposures, constructs a Lyα template from the 2007–2023 orbital baseline, and makes electronic data available. The high-latitude, outer-solar-system vantage point removes geocoronal and zodiacal foregrounds that have plagued earlier measurements. These strengths make the paper valuable regardless of the final verdict. However, the central first-detection claim depends on two assumptions that need explicit support: that the Box spectrum in the 912–1100 Å band is not dominated by red-wing leakage from longer wavelengths, and that a single GALEX-based scale factor, calibrated at 1400–1700 Å, applies across the entire Alice spectrum including the KBr-coated short-wavelength region.","major_comments":[{"comment":"The first-firm-detection claim at 912–1100 Å is not yet supported because the analysis does not address the spectral purity of the Box measurement. The Box has a FWHM of 172 Å (§2.1), nearly the full 188 Å width of the claimed band, and §4.4 states that the detector region was integrated down to 850 Å to capture the short-wavelength signal. Without a line-spread-function deconvolution or an explicit leakage estimate, the 912–1100 Å band will include red-wing counts from the 1100–1350 Å region (blanked in Fig. 15) and potentially from 1400–1800 Å, where the measured offset is 264 ± 24 photon units. The paper itself cautions that the decrease in offsets from 1100 to 912 Å 'may be an artifact of the Box line width.' The quoted 221 ± 11 photon units includes only the statistical errors from the fit and template subtraction, not an LSF modeling error. The Stem measurement in the same band, 172 ± 48 photon units, is far less precise and cannot by itself establish a firm detection. Please quantify the red-wing contribution using the measured LSF and the observed spectrum at λ > 1100 Å, or deconvolve the spectrum; unless this is done, the first-detection claim is not load-bearing.","section":"§4.4 / §2.1"},{"comment":"The absolute calibration of the 912–1100 Å band rests on an assumption that is stated but not validated. Section 4.2 derives a single scale factor from the correlation between Alice 1400–1700 Å counts and GALEX FUV background, and then 'rescaled the Alice spectra by these factors, assuming that a single scale factor applies over the entire Alice spectrum.' The Alice detector has a split coating of KBr (520–1180 Å) and CsI (1250–1870 Å) (§2.1). If the relative sensitivity of the two coatings has drifted since ground calibration, or if the diffuse-source calibration differs between the two coatings, the 1000 Å offset of 221 ± 11 photon units would shift by an amount not captured by the quoted uncertainty. Please provide evidence that the short-wavelength response tracks the 1400–1700 Å response, for example from stellar observations at 912–1100 Å or from a stability analysis of the KBr/CsI boundary region, or propagate a conservative systematic uncertainty into the 1000 Å offset.","section":"§4.2"},{"comment":"The reported EUV limit of 3.2 ± 3.0 photon units at 600–800 Å is presented as an absolute surface brightness, but it is converted using the same GALEX-derived calibration that is only established for 1400–1700 Å. The text does not describe any calibration check for the 520–912 Å KBr-coated region, which is the part of the detector most likely to have a different sensitivity history. Either provide a validation for the short-wavelength calibration or present the EUV result as an upper limit in count-space rather than in photon units, with the calibration uncertainty included.","section":"§4.3"}],"minor_comments":[{"comment":"Please clarify the direction of the rescaling: the fit gives Alice = 0.71 × GALEX for the Box, so the correction factor to apply to Alice spectra should be 1/0.71, yet the text says 'rescaled the Alice spectra by these factors' without specifying whether 'factors' means the slope or its inverse.","section":"§4.2"},{"comment":"The caption of Fig. 8 writes the subtracted component as D2023 + T × L2007, while Eq. (3) defines CUVB = (S − D) − L ∗ T; please reconcile the notation for the template scaling so that L and T are not confused with each other.","section":"§3.3 / Fig. 8"},{"comment":"The 912–1100 Å STEM row reports a slope of 2214 ± 1694 photon units mag−1, a relative uncertainty of ~76%; please state explicitly that the Stem is too noisy to constrain the slope and that all quantitative conclusions rely on the Box.","section":"Table 7"},{"comment":"The text says the offsets are 'close to flat between 1400 – 1800 ˚A at a level of about 290 photon units,' but Table 7 quotes 264 ± 24 photon units for 1400–1700 Å; please specify whether the 290 value refers to a different wavelength range or a different estimate.","section":"§4.4"},{"comment":"The error propagation description adds dark, sky, and template errors in quadrature, but it does not state how the template normalization uncertainty (the scaling by the Lyα line ratio L2007/L2023) is incorporated into the template error; please add one sentence describing the scaling uncertainty.","section":"§3.3.1"}],"recommendation":"major_revision","confidential_remarks":"This is an important observational result that will attract wide attention if the calibration and spectral-isolation concerns are resolved. The paper's careful dark handling and Lyα-template construction are strengths. The two load-bearing issues are the red-wing leakage into the 912–1100 Å Box band and the single-scale-factor calibration across the KBr/CsI coating boundary; both are addressable in revision. I see no concerns about data provenance or citation practices."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague—\n\nRead the New Horizons Alice CUVB paper. Bottom line: the 1500 Å result is solid and consistent with GALEX; the 912–1100 Å 'first firm detection' is a real measurement but the quoted 221 ± 11 photon units does not include the systematic from the Box's line-spread function, and that is the right thing to push on.\n\nWhat the paper does well: the data are from 57 AU, outside most interplanetary Lyα and zodiacal foregrounds. The analysis is unusually careful about dark current—interleaved dark exposures, tracking slow drift—and the Lyα scattering template built by differencing 2007 and 2023 observations of the same sky is clever. The correlation with Planck E(B−V) is clean, and the 264 ± 24 photon units at 1500 Å matches the GALEX-based offsets. The EUV upper limit below 912 Å is a useful new constraint. This is careful, well-documented work, and the authors are honest about their own caveats.\n\nThe soft spot: the first-detection claim at 1000 Å. The Box has 172 Å FWHM for diffuse sources, and the 912–1100 Å band is only 188 Å wide. The paper integrates down to 850 Å to capture the signal, but never deconvolves the line-spread function, so red-wing leakage from the longer-wavelength continuum (where the offset is 264 photon units) can leak into the short-wavelength band. The authors themselves note that the decline in offsets from 1100 to 912 Å 'may be an artifact of the Box line width.' The higher-resolution Stem gives 172 ± 48 photon units—consistent but far too imprecise to anchor the claim. So the headline number is carried by a measurement whose spectral purity is not demonstrated. A second issue: the absolute scale is set by rescaling the full Alice spectrum with a single factor from a GALEX 1400–1700 Å comparison; if the short-wavelength response drifted relative to the long-wavelength response, the 1000 Å offset shifts by more than the stated ±11.\n\nI don't think this is fatal. The 1500 Å excess, about half unexplained, has support from GALEX and other experiments. The 1000 Å detection is plausibly real, but the current analysis does not quite secure it. A referee should ask for an LSF/red-wing leakage estimate or a deconvolution, and for the scale-factor systematic to be folded into the quoted uncertainty.\n\nThis is for anyone working on the UV background, the extragalactic background light, halo gas, or future FUV missions. It deserves a serious referee; the unique vantage point and careful reduction are worth the time. I would engage with it, but I would treat the 1000 Å number as provisional until the LSF question is addressed.\n\nRecommendation: send to peer review with a request for a concrete LSF systematics estimate.","headline":"The 1500 Å result is solid; the 1000 Å 'first detection' needs an LSF systematics estimate before I'd trust the quoted uncertainty.","tokens_in":24068,"tokens_out":4532,"would_cite":true,"duration_ms":40582,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"New Horizons measurements from 57 AU give the first firm measurement of the cosmic ultraviolet background in the 912–1100 Å band and find that roughly half of the high-latitude UV background is unidentified.","keywords":["cosmic ultraviolet background","diffuse radiation","New Horizons Alice","high Galactic latitude","Lyman limit","two-photon emission","interstellar reddening","GALEX cross-calibration"],"falsifier":"To test the first-detection claim, one could observe a hot white dwarf or B star with a model-atmosphere spectrum through Alice's short-wavelength (KBr) and long-wavelength (CsI) channels and compare the 912–1100 Å count rate with the model prediction; if the single GALEX-derived scale factor is wrong for the short-wavelength channel by more than the quoted errors, the $221 \\pm 11$ photon units offset at 1000 Å would shift and the firm detection would be called into question.","tokens_in":22825,"feed_emoji":"🔭","tokens_out":16593,"duration_ms":137317,"temperature":0.7,"pith_summary":"Cosmic ultraviolet background measurements from Earth orbit are contaminated by airglow and by scattered interplanetary Lyman-$\\alpha$. This paper reports observations with the Alice spectrograph on New Horizons from 57 AU, where those foregrounds are largely gone, of 25 high-latitude fields ($|b|>40^\\circ$). The background is linearly correlated with the Planck $E(B-V)$ reddening, with zero-reddening offsets of $221 \\pm 11$ photon units at 1000 Å and $264 \\pm 24$ photon units at 1500 Å ($4.4 \\pm 0.2$ and $5.3 \\pm 0.5$ nW m$^{-2}$ sr$^{-1}$). The 1000 Å offset is claimed as the first firm detection in the 912–1100 Å window, where the prior result was only an upper limit. Known sources explain about half of the signal, leaving unidentified excesses of $133 \\pm 17$ photon units at 1000 Å and $103 \\pm 31$ photon units at 1500 Å; nothing is detected shortward of the Lyman limit.","feed_headline":"Cosmic UV background in 912–1100 Å measured for first time","feed_subtitle":"New Horizons measures a 221-photon-unit offset at 1000 Å that no known source explains.","key_machinery":"The carrier of the argument is the data-processing sequence that turns Alice raw counts into a clean background spectrum, built around two foreground subtractions. Dark counts from the spacecraft's radioisotope thermoelectric generator are measured with door-closed exposures interleaved with the sky exposures. The dominant contaminant, scattering of the intense interplanetary Lyman-alpha line across the detector, is removed with a template formed by differencing Alice observations of the same sky region made in 2007 at about 8 AU and in 2023 at about 57 AU, scaled by the Lyman-alpha line counts; the solar Lyman-alpha signal fell by a factor of 3.5 over that interval, so the difference isolates the scattering function. Because aperture-filling diffuse calibration is difficult, the 1400–1700 Å Alice band is cross-calibrated against GALEX diffuse observations, and a single scale factor is applied to the whole Alice spectrum. The residual after these subtractions, plotted against the mean Planck $E(B-V)$ in each aperture, yields the linear fits and zero-reddening offsets.","core_discovery":"The paper's central claim is that the cosmic ultraviolet background at high Galactic latitudes has a substantial component not explained by any known source, and that this component can now be measured in the previously inaccessible 912–1100 Å band. From 25 fields observed with the Alice spectrograph at 57 AU, after subtracting dark counts and a scattering template for interplanetary Lyman-$\\alpha$, the CUVB surface brightness is linear in the Planck $E(B-V)$: at 1000 Å the best fit is $2994 \\pm 446$ photon units per magnitude of $E(B-V)$ plus an offset of $221 \\pm 11$ photon units; at 1500 Å it is $6723 \\pm 909$ photon units per magnitude plus $264 \\pm 24$ photon units, where photon units are photons cm$^{-2}$ s$^{-1}$ sr$^{-1}$ Å$^{-1}$. The 1000 Å offset is the first firm measurement of the zero-reddening offset between 912 and 1100 Å; the only earlier measurement, from the Voyager ultraviolet spectrometers, was an upper limit near 200 photon units. Adding the integrated light of unresolved galaxies, faint stars, O VI and C IV line emission, and two-photon emission accounts for roughly half the offsets, leaving $133 \\pm 17$ photon units at 1000 Å and $103 \\pm 31$ photon units at 1500 Å with no identified source. In the 600–800 Å band the measured surface brightness is $3.2 \\pm 3.0$ photon units, so no background is detected below the Lyman limit.","pith_inferences":["A natural next test, not performed in the paper, is to replace the single GALEX-derived scale factor with a wavelength-dependent calibration anchored to stellar model atmospheres; if the 1000 Å offset moves by more than its stated error, the first-detection claim would rest on a calibration artifact.","If the residual excess is a real astrophysical component, cross-correlating the 1000 Å residual map with galaxy redshift surveys, X-ray maps, and far-infrared dust maps would show whether it is Galactic halo emission or genuinely extragalactic; the paper already cites a clustering analysis that argues against an extragalactic origin for a comparable 1500 Å monopole.","The combination of a substantial 912–1100 Å offset with a null below 912 Å implies that the unknown emitter produces photons just longward of the Lyman limit but not ionizing photons; warm partially ionized gas or a decaying-particle source would make concrete predictions that future short-wavelength UV observations could test.","If a future deep-space platform observes the same 912–1100 Å window with an independently calibrated spectrograph, the comparison would be a clean check of whether the 221 photon units offset is a property of the sky or of Alice's calibration."],"forward_implications":["The 912–1100 Å zero-reddening offset becomes a measured number, $221 \\pm 11$ photon units, rather than an upper limit, so models of the extragalactic UV background must now reproduce it.","After accounting for galaxies, stars, line emission, and two-photon emission, an excess of $133 \\pm 17$ photon units at 1000 Å and $103 \\pm 31$ photon units at 1500 Å remains, indicating an unidentified isotropic component of the high-latitude UV sky.","The null detection below the Lyman limit, $3.2 \\pm 3.0$ photon units in the 600–800 Å band, constrains the metagalactic ionizing background and implies that the local EUV field seen by earlier missions is dominated by a few nearby stars.","Because the 1400–1800 Å Alice and GALEX offsets agree to $-23 \\pm 24$ photon units, the paper concludes that 1300–1800 Å background measurements are feasible from low Earth orbit, whereas the 912–1100 Å band requires an outer-solar-system vantage point where Lyman-alpha scattering is weak."],"supporting_citations":[{"why":"supplies the only prior 912–1150 Å measurement, an upper limit that the new 1000 Å detection supersedes.","marker":"Holberg (1986)"},{"why":"provides the adopted galaxy-count extragalactic background, 73 ± 16 photon units at 1500 Å, the main known-source component.","marker":"Driver et al. (2016)"},{"why":"supplies the EBL spectrum used to estimate the galaxy contribution of less than 10 photon units at 1000 Å.","marker":"Koushan et al. (2021)"},{"why":"quantifies the two-photon emission contribution of about 23 photon units and the atmospheric contamination of earlier low-Earth-orbit measurements.","marker":"Kulkarni (2022)"},{"why":"provides the O VI line emission estimate used in the component budget for the 912–1100 Å band.","marker":"Shelton et al. (2001)"},{"why":"supplies the stellar-population synthesis model used to estimate unresolved faint-star contributions at 1000 and 1500 Å.","marker":"Girardi et al. (2005)"},{"why":"supplies the GALEX point-source catalog used to count resolved hot stars that contribute to the offsets.","marker":"Bianchi et al. (2018)"},{"why":"provides the GALEX-based correlation of CUVB with reddening and the 230–290 photon units offsets that the 1500 Å result confirms.","marker":"Akshaya et al. (2019)"},{"why":"defines the field selection, far-infrared DGL estimator, and survey design that the Alice fields were chosen to parallel.","marker":"Postman et al. (2024)"}],"fun_headline_variants":["New Horizons reveals mysterious excess UV emission from deep space","Half of cosmic UV background at 1000 Å has unknown source","First UV background offset at 1000 Å measured by New Horizons","Excess UV light in cosmos: first measurement at 1000 Å","New Horizons finds unexplained UV glow at high galactic latitudes"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that a single scale factor, derived from a 1400–1700 Å comparison with GALEX, applies unchanged across the entire Alice spectrum, including the 912–1100 Å band that uses a different detector coating.","fun_headline_variants_meta":{"raw":{"variants":["New Horizons reveals mysterious excess UV emission from deep space","Half of cosmic UV background at 1000 Å has unknown source","First UV background offset at 1000 Å measured by New Horizons","Excess UV light in cosmos: first measurement at 1000 Å","New Horizons finds unexplained UV glow at high galactic latitudes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001045,"raw_usage":{"total_tokens":4532,"prompt_tokens":1223,"completion_tokens":3309,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":839,"completion_tokens_details":{"reasoning_tokens":3220}},"tokens_in":839,"tokens_out":3309,"duration_ms":19220,"temperature":1.0,"reasoning_tokens":3220,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T22:42:17.156908+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"To test the first-detection claim, one could observe a hot white dwarf or B star with a model-atmosphere spectrum through Alice's short-wavelength (KBr) and long-wavelength (CsI) channels and compare the 912–1100 Å count rate with the model prediction; if the single GALEX-derived scale factor is wrong for the short-wavelength channel by more than the quoted errors, the $221 \\pm 11$ photon units offset at 1000 Å would shift and the firm detection would be called into question.","supporting_citations":[{"cited_title":"2018, Ap&SS, 363, 56, doi: 10.1007/s10509-018-3277-2","cited_arxiv_id":null,"evidence_quote":"supplies the GALEX point-source catalog used to count resolved hot stars that contribute to the offsets."}],"review_version":1}