Hypothesis

Five astronomical surveys totaling 6.9 million sources -- spanning radio continuum and optical/IR spectroscopy -- show number-count dipole directions that cluster within a mean pairwise separation of 23.7 degrees, where random orientations predict 90 +/- 12 degrees. Monte Carlo testing (100,000 trials) yields p < 10^-5 for the directional clustering (4.4 sigma). This suggests a genuine preferred direction in the large-scale matter distribution.

Context and Prior Art

The observation that cosmic dipole directions cluster is well-established. Secrest et al. (2025, Reviews of Modern Physics, arXiv:2505.23526) reviewed the cosmic dipole anomaly at >5 sigma significance across multiple surveys. Bohme et al. (2025, PRL, arXiv:2509.16732) combined NVSS+RACS+LoTSS for a 5.4 sigma radio-only detection. This analysis contributes a new multi-survey directional consistency test using our own re-analysis of publicly available catalogs.

Survey Data

All dipole directions extracted via spherical harmonic decomposition of number counts in HEALPix pixels from publicly available catalog data.

| Survey | Type | N Sources | Dipole Direction (l, b) | Internal Consistency | |--------|------|-----------|------------------------|---------------------| | DESI LRG | Spectroscopic galaxies | 2,400,000 | (222.9, 70.9) | z-bin scatter: 1.5 deg (3.1 sigma) | | DESI QSO | Spectroscopic quasars | 1,300,000 | (234.2, 53.7) | z-bin scatter: 0.7 deg (3.1 sigma) | | FIRST | 1.4 GHz radio | 946,432 | (221.2, 72.9) | Flux-split: 3.4 deg (3.1 sigma) | | SDSS | Spectroscopic galaxies | 500,000 | (203.7, 67.6) | z-bin scatter: 6.5 deg (2.2 sigma) | | NVSS | 1.4 GHz radio | 1,773,484 | (239.1, 27.8) | Flux-split: 3.0 deg (3.2 sigma) |

All directions in Galactic coordinates (degrees). Total: 6.92 million sources.

Key Results

Directional Clustering

The 10 pairwise angular separations have a mean of 23.7 degrees. In 100,000 Monte Carlo trials of 5 random directions, p < 10^-5 (4.4 sigma).

NVSS Offset

NVSS (b=27.8) is offset ~44-46 degrees from the other four surveys, all of which have b > 50 degrees. This may reflect the different source population (bright radio AGN) or known NVSS systematics. Excluding NVSS, the remaining 4 surveys cluster with mean separation 12.6 degrees (4.3 sigma).

Internal Stability

Each survey's dipole direction is internally consistent:

• DESI LRG: direction stable to 1.5 degrees RMS across 7 redshift bins (z = 0.4 to 1.1)
• DESI QSO: direction stable to 0.7 degrees RMS across 8 redshift bins (z = 0.8 to 3.0)
• NVSS: bright/faint flux split gives 3.0 degree separation (3.2 sigma internal consistency)
• FIRST: bright/faint flux split gives 3.4 degree separation (3.1 sigma)

Consensus Direction

Weighted mean across all 5 surveys: (l = 230, b = 58)

This is offset 27.7 degrees from the CMB kinematic dipole (l=264, b=48). The offset does not rule out a kinematic origin but suggests the signal may have an additional component.

Important Caveats

1. None of these surveys are full-sky. DESI, SDSS, and FIRST cover primarily the North Galactic Cap visible from their respective facilities. Partial sky coverage can bias dipole direction estimates, particularly toward high galactic latitudes.
2. DESI LRG and FIRST dipoles are suspiciously similar -- both at (l ~ 221, b ~ 72), separated by only 2.1 degrees. This may partly reflect overlapping sky coverage rather than independent confirmation.
3. Dipole fitting methodology matters. Different estimators (pixel counting, spherical harmonics, maximum likelihood) can yield different directions for partial-sky surveys. We used a single method (spherical harmonics in HEALPix) for consistency, but this choice introduces methodological dependence.
4. Fisher's combined significance (6.2 sigma from combining internal p-values) assumes the 5 internal consistency tests are independent. While they test different data (different surveys, different splits), there may be residual correlations from shared sky regions. The 4.4 sigma from directional clustering alone is more robust.
5. We report dipole direction convergence, not amplitude anomaly. Whether the dipole amplitudes are anomalously large is separately debated (MeerKAT vs. NVSS/CatWISE disagreement).

Notable Sub-Result

In an earlier analysis of a DESI DR1 subset (296,428 LRGs and 240,865 QSOs), LRG and QSO dipoles aligned to 0.33 degrees (p = 4 x 10^-6, 4.5 sigma). The full DESI LSS catalog (3.7 million sources) yields a larger LRG-QSO separation of 17.9 degrees, likely reflecting the sensitivity of dipole estimation to sky coverage and sample selection. The sub-result remains interesting but should be interpreted cautiously.

Falsification Criteria

This hypothesis is falsified if:

• Full DESI data releases show dipole directions shifting by > 30 degrees from current values
• Euclid or LSST full-sky surveys find number-count dipoles pointing > 45 degrees from our consensus direction
• The clustering is demonstrated to arise from a common systematic (e.g., Galactic extinction, survey footprint geometry)
• Independent re-analysis with different dipole estimators yields inconsistent directions

Methodology Note

All analyses used real, publicly available catalog data (DESI DR1 LSS v1.5, NVSS via VizieR, FIRST VIII/92, SDSS DR17). Code and processed results are available for reproduction. Monte Carlo significance was computed by comparing observed pairwise separations against random direction ensembles on the sphere.

Key references: Secrest et al. 2025 (arXiv:2505.23526); Bohme et al. 2025 (arXiv:2509.16732); Wagenveld et al. 2024 (arXiv:2408.16619); Aluri et al. 2022 (arXiv:2207.05765).