⚛️ general relativity

Eppur non si trovano: Comments on the Primordial Black Hole Limits in the Galactic Halo

This paper refutes the claims made by Hawkins & Garcia-Bellido regarding the OGLE microlensing survey, reaffirming that primordial black holes and other compact objects cannot constitute a substantial fraction of the Milky Way's dark matter halo due to the lack of detected microlensing events.

Original authors: P. Mróz, A. Udalski, M. K. Szymański, I. Soszyński, Ł. Wyrzykowski, P. Pietrukowicz, S. Kozłowski, R. Poleski, J. Skowron, D. Skowron, K. Ulaczyk, M. Gromadzki, K. Rybicki, P. Iwanek, M. Wrona, M. Rat

Published 2026-01-27

📖 6 min read🧠 Deep dive

CC BY 4.0

Original authors: P. Mróz, A. Udalski, M. K. Szymański, I. Soszyński, Ł. Wyrzykowski, P. Pietrukowicz, S. Kozłowski, R. Poleski, J. Skowron, D. Skowron, K. Ulaczyk, M. Gromadzki, K. Rybicki, P. Iwanek, M. Wrona, M. Ratajczak

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

The Big Picture: A Cosmic "Missing Persons" Case

Imagine the Milky Way galaxy as a giant city. We know this city has a lot of "invisible weight" (dark matter) holding it together, but we can't see it. For decades, astronomers had a theory: maybe this invisible weight is made of billions of tiny, invisible black holes (called Primordial Black Holes) floating around in the dark halo surrounding our galaxy.

If this theory were true, these invisible black holes would act like invisible magnifying glasses. As they drifted in front of distant stars, they would briefly brighten the star's light, creating a specific "blink" pattern.

The Conflict:

The Old Theory (MACHO Project): In the late 90s and early 2000s, a team called MACHO looked at the sky and claimed to see about 13 to 17 of these "blinks." They said, "Aha! We found them! The dark matter is made of these black holes."
The New Reality (OGLE Project): A newer, much larger team called OGLE watched the same patch of sky for 20 years. They looked at 80 million stars. They only found 13 blinks total, and most of those were likely caused by normal stars, not black holes. They concluded: "We can't find the black holes. They probably don't exist in large numbers."

Recently, two critics (Hawkins & García-Bellido) published a paper saying, "Wait a minute! The OGLE team is bad at looking. They are missing the black holes because their tools are too clumsy. The MACHO team was right."

This paper is the OGLE team's response. They are saying: "No, you are wrong. Our tools are actually much better, and the critics are misinterpreting the data. The black holes are still missing."

The Arguments: Why the Critics Are Wrong

The authors break down the critics' complaints one by one, using simple logic and better data.

1. The "Blurry Camera" vs. The "Sharp Camera"

The Critic's Claim: The OGLE team's photos are too blurry. In crowded star fields, stars overlap, making it hard to tell if a star is just blinking (a variable star) or if it's being magnified by a black hole. They say OGLE's data is "noisy."

The OGLE Rebuttal:
Think of the MACHO team's method like trying to listen to a single person speak in a noisy room using an old, low-quality microphone. You hear a lot of background chatter.
The OGLE team uses a technique called Difference Image Analysis (DIA). Imagine taking a photo of the room, then taking another photo a few days later. You subtract the first photo from the second.

Result: All the static objects (the furniture, the walls, the non-blinking stars) disappear. You are left with a "ghost image" showing only the things that changed.
The Proof: The authors show that their "noise" is actually 2 to 3 times lower than MACHO's. Their "microphones" are crystal clear. The critics' claim that OGLE is missing events because of noise is like saying a high-definition camera is worse than a blurry one.

2. The "Imposter" Stars

The Critic's Claim: The MACHO team found 13–17 events, and the OGLE team only found 13. The critics say OGLE is just being too picky and throwing away good data.

The OGLE Rebuttal:
The authors went back and re-analyzed the original MACHO "blinks" using their new, sharp "Difference Image" technique. They discovered that five of the MACHO "blinks" were actually imposters.

The Analogy: Imagine a police lineup. MACHO pointed at five people and said, "These are the thieves!" OGLE re-examined them and realized, "Wait, these five are actually just actors rehearsing a play. They have a pattern of blinking that repeats every few days."
The Discovery: Two of these "imposters" were found to be a new type of exploding star system (called "millinovae") that flashes repeatedly. Three others were just pulsing stars.
The Result: If you remove these five fakes from the MACHO list, their "evidence" for black holes drops by 40%. The remaining events are so few that they don't support the idea that the galaxy is full of black holes.

3. The "Blue Star" Bias

The Critic's Claim: OGLE is biased against finding black holes in front of blue stars (hot, young stars). They claim OGLE only sees red stars, while MACHO saw blue ones.

The OGLE Rebuttal:
The authors checked the "color" of the stars in their data. They showed a chart proving that OGLE sees blue stars just fine. The confusion came because the MACHO photos were so blurry that a bright blue star next to a dim star made the whole group look blue. When OGLE looked with their sharp "Difference Image" glasses, they saw the true colors. The bias didn't exist; it was just a measurement error in the old data.

4. The "Timing" Argument

The Critic's Claim: OGLE doesn't look at the sky often enough (every 3–10 days). They say they might miss short blinks that happen between observations.

The OGLE Rebuttal:
The authors checked their own schedule. In the most important areas of the sky, they look every 2 days on average. This is fast enough to catch the events they are looking for. They also ran computer simulations showing that even if they missed a few, it wouldn't change their main conclusion: there simply aren't enough black holes to explain the MACHO results.

The Final Verdict

The paper concludes with a strong statement: "Eppur non si trovano" (And yet they are not found).

Despite 20 years of watching 80 million stars with the best technology available, and despite re-examining the old data with new tools, the evidence for a galaxy filled with primordial black holes is still missing.

The MACHO "hits" were likely a mix of real events and many false alarms (variable stars).
The OGLE "misses" were actually a success: they successfully filtered out the noise and found that the universe is much emptier of these specific black holes than previously thought.

The authors argue that if the entire dark matter halo were made of these black holes, they should have seen hundreds of events. Instead, they saw only a handful, which can be explained by normal stars. The "invisible black holes" remain invisible, suggesting they are not the main ingredient of the universe's dark matter.

Technical Summary: "Eppur non si trovano: Comments on the Primordial Black Hole Limits in the Galactic Halo"

Problem Statement
This paper addresses a recent critique by Hawkins & García-Bellido (2025) regarding the results of the 20-year OGLE photometric monitoring of the Magellanic Clouds. The OGLE survey found a scarcity of gravitational microlensing events, implying that primordial black holes (PBHs) and other compact objects in the mass range of $10^{-8}$ to $10^3 M_\odot$ cannot constitute a substantial fraction of the Milky Way's dark matter halo. Hawkins & García-Bellido (2025) challenged these findings, suggesting that OGLE's methodology suffers from biases (e.g., against blue sources, poor cadence, and photometric quality in crowded fields) that led to a significant undercount of events, thereby invalidating the constraints on PBHs. This paper aims to refute these claims by demonstrating that the arguments presented by Hawkins & García-Bellido lack a solid basis or are factually incorrect.

Methodology
The authors employ a multi-faceted approach to re-evaluate the data and critique the opposing arguments:

Re-analysis of MACHO Data: The authors applied modern Difference Image Analysis (DIA) techniques to the original MACHO project images. Unlike the MACHO team's original Point Spread Function (PSF) fitting, DIA subtracts a high-quality reference image from science images to isolate variable sources, effectively removing contamination from constant stars in dense fields.
Characterization of MACHO Candidates: Using the new DIA light curves, the authors re-examined the 13–17 microlensing candidates reported by Alcock et al. (2000). They cross-matched these candidates with OGLE databases to obtain precise magnitudes and colors, correcting for blending effects that may have distorted the original MACHO measurements.
Statistical Validation: The authors compared the observed distributions of microlensing event parameters (impact parameter $u_0$ and peak time $t_0$ ) against simulated distributions derived from their detection efficiency models. They utilized the Kolmogorov–Smirnov test to verify the consistency between observed and expected data.
Model Verification: The authors recalculated the microlensing optical depth ( $\tau$ ) for the Large Magellanic Cloud (LMC) using various Milky Way halo models, including the Jiao et al. (2023) and Ou et al. (2024) models cited by the critics, to test the sensitivity of the PBH limits to the assumed dark matter distribution.

Key Contributions and Results

Re-evaluation of MACHO Candidates: The DIA reduction of MACHO images revealed that at least five of the original MACHO candidates (MACHO-LMC-07, 08, 18, 23, 27) are not genuine microlensing events.
- Two objects (LMC-07 and LMC-23) exhibit repeating outbursts and are identified as "millinovae," a new class of cataclysmic variables.
- Three objects (LMC-08, LMC-18, LMC-27) show periodic photometric variability consistent with binary stars rather than microlensing.
- Removing these five variable objects reduces the MACHO-derived microlensing optical depth by nearly 40%, from $1.2 \times 10^{-7}$ to $0.74 \times 10^{-7}$ , bringing it closer to OGLE and EROS limits.
Refutation of OGLE Biases: The authors systematically dismantle the specific criticisms raised by Hawkins & García-Bellido:
- Photometric Quality: DIA photometry in OGLE yields a root mean square (rms) scatter 2.45 to 3.08 times lower than MACHO PSF photometry, making OGLE superior in crowded fields, not inferior.
- Achromaticity: The authors argue that achromaticity is a second-order condition. With 20 years of data, repeating outbursts can be identified regardless of color information. Furthermore, DIA removes blending effects that could cause spurious chromaticity.
- Cadence: The median cadence in central LMC fields is 2.0–3.0 days, comparable to MACHO, and sufficient for detecting microlensing events.
- Source Color Bias: Analysis of the color-magnitude diagram (CMD) using OGLE data shows that blue main-sequence sources are present in the OGLE sample; the claim that OGLE misses them is incorrect.
- Finite-Source Effects: Calculations show that for typical halo lenses ( $1 M_\odot$ ), finite-source effects are negligible ( $\rho < 10^{-2}$ ) for the vast majority of source stars.
Detection Efficiency Validation: The observed distributions of impact parameters ( $u_0$ ) and peak times ( $t_0$ ) for the 13 OGLE events match the simulated distributions with high statistical significance (p-values of 0.983 and 0.938, respectively). This confirms that the detection efficiency calculations used by Mróz et al. (2024a) are robust.
Halo Model Independence: Recalculations using alternative Milky Way halo models (Jiao et al., 2023; Ou et al., 2024) show that the expected number of microlensing events changes by only a few percent compared to the fiducial Cautun et al. (2020) model. Thus, the conclusion that PBHs cannot comprise a large fraction of dark matter remains robust regardless of the specific halo model used.

Significance and Conclusions
The paper concludes that the Hawkins & García-Bellido (2025) critique is unfounded. The discrepancy between the MACHO results (early 2000s) and the OGLE results (2001–2020) is largely explained by the superior photometric techniques of OGLE (DIA vs. PSF) and the identification of variable stars that were previously misclassified as microlensing events.

The authors reiterate that despite extensive searches covering nearly 80 million stars over two decades, only 13 genuine microlensing events were found in the direction of the LMC. This number is consistent with expectations from known stellar populations (self-lensing and disk lenses) but far too low to support the hypothesis that the Milky Way dark matter halo is composed of primordial black holes or other compact objects in the mass range of $10^{-8}$ to $10^3 M_\odot$ . The paper asserts that, as the title suggests, "Eppur non si trovano" (And yet they are not found): compact objects in the dark halo remain undetected.