Asymmetry Between Galaxies With Clockwise Handedness and Counterclockwise Handedness

Abstract

While it is clear that spiral galaxies can have different handedness, galaxies with clockwise patterns are assumed to be symmetric in all of their other characteristics to galaxies with counterclockwise patterns. Here we use data from SDSS DR7 to show that photometric data can distinguish between clockwise and counterclockwise galaxies. Pattern recognition algorithms trained and tested using the photometric data of a clean manually crafted dataset of 13,440 spiral galaxies with z < 0.25 can predict the handedness of a spiral galaxy in ∼64% of the cases, significantly higher than mere chance accuracy of 50% (P < 10 −5 ). Experiments with a different dataset of 10,281 automatically classified galaxies showed similar results of ∼65% classification accuracy, suggesting that the observed asymmetry is consistent also in datasets annotated in a fully automatic process, and without human intervention. That shows that the photometric data collected by SDSS is sensitive to the handedness of the galaxy. Also, analysis of the number of galaxies classified as clockwise and counterclockwise by crowdsourcing shows that manual classification between spiral and elliptical galaxies can be affected by the handedness of the galaxy, and therefore galaxy morphology analyzed by citizen science campaigns might be biased by the galaxy handedness. Code and data used in the experiment are publicly available, and the experiment can be easily replicated.

Photometric Asymmetry Between Clockwise and Counterclockwise Spiral Galaxies in SDSS

“…The experiment using the dataset of galaxies that were annotated in a fully automatic process provided a comparable handedness prediction accuracy of ∼65%. The analysis also revealed that several different photometric measurements such as the SDSS 'Stokes U' parameter exhibit a statistically significant difference between clockwise and counterclockwise galaxies (Shamir 2016).…”

“…Linear regression is applied across the vertical lines of the radial intensity plot for each group of peaks, and the sign of the slopes reflect the direction of the arm, therefore determining the handedness of the galaxy. The algorithm is described in details and numerous examples in (Shamir 2011b;Hoehn & Shamir 2014;Shamir 2012), and the process of galaxy classification is described in (Shamir 2016).…”

“…While galaxies with clockwise handedness are expected to be symmetric to galaxies with counterclockwise handedness, previous experiments (Shamir 2016) have shown that clockwise galaxies imaged by Sloan Digital Sky Survey (York et al 2000) are photometrically different from galaxies with counterclockwise patterns. The experiment was done by separating a population of spiral galaxies into clockwise and counterclockwise galaxies, and then collecting the photometric information of each galaxy from SDSS Catalog Archive Server (CAS).…”

“…Similarly to (Shamir 2016), the spiral galaxies were separated by their handedness to clockwise and counterclockwise galaxies using the Ganalyzer algorithm (Shamir 2011b,a), which converts each galaxy image to its radial intensity plot, and then applies automatic peak detection to find and group the peaks along the horizontal line of the radial intensity plot (Shamir 2011b). Since the galaxy arms are brighter than nonarm pixels at the same radial distance, the peaks are expected to identify the arms.…”

“…Land et al (2008) used a large dataset of galaxies annotated by crowdsourcing, showing that after correcting for the substantial human bias the number of galaxies with clockwise handedness was higher than the number of counterclockwise galaxies, but the difference was not statistically significant (Land et al 2008). A more recent analysis using 13,440 automatically classified galaxies (Shamir 2016) showed a higher number of galaxies with clockwise handedness compared to galaxies with counterclockwise handedness, which is also in 2 Shamir agreement with the higher number of Galaxy Zoo (Lintott et al 2011) clockwise galaxies observed in (Shamir 2012). On the other hand, the dataset of 10,281 galaxies used in (Shamir 2016) showed no statistically significant preference (Shamir 2016).…”