Towards a Multi-Dimensional Stellar Classification System

J. Stock, R. Agostinho, J. A. Rose, Z. Marín, Rafael MolinaOpen 1997 · DOI

Abstract: Practically all the MK spectral types known today are the result of visual inspection of photographic spectra. This includes the monumental work by Nancy Houk. Two questions related to the MK system remained unanswered all along, namely 1) How accurate are the luminosities and intrinsic colors derived from these types? And 2) Can the classification be done on the basis of measured spectroscopic criteria such as equivalent widths? A third question must also be raised, namely 3) Should we not rather classify in a system with more than two dimensions? With the Parallax Catalogue of Also now techniques are available to produce measured spectroscopic criteria in large quantities. The first steps in this direction will be discussed here.

Introduction

For decades, the Morgan-Keenan (MK) system has served as the standard for classifying stars based on their spectra. However, this system relies heavily on the qualitative "visual inspection" of photographic plates. This paper identifies three major gaps in the traditional approach: the inherent subjectivity that hinders scalability for modern surveys, the uncertain accuracy of physical parameters derived from qualitative types, and the oversimplification of a two-dimensional (temperature and luminosity) classification system. The authors propose a shift toward a quantitative, multi-dimensional system to more accurately characterize stellar physics.

Methodology

The proposed methodology moves away from subjective assessment toward quantitative measurements of spectroscopic criteria, specifically Equivalent Widths (EW) of absorption lines. To ensure robustness across different instruments without requiring absolute flux calibration, the authors utilize a pseudo-continuum technique, interpolating a baseline between intensity peaks surrounding a spectral feature. This approach remains effective even at lower resolutions, focusing on strong features like the Hydrogen-delta line to maintain applicability for fainter stars. By measuring a suite of 12 different spectral lines, the authors define a star's position within a multi-parameter space rather than assigning a single discrete class.

Data and Calibration

The study utilizes high-precision parallax data from the HIPPARCOS mission and the General Catalogue of Trigonometric Stellar Parallaxes to establish a reliable "ground truth" for stellar distances and luminosities. This calibration backbone is supplemented by slit spectra of several hundred bright stars from the University of Hawaii’s 2.2-meter telescope and objective-prism plates for nearly 3,000 stars from the Curtis Schmidt telescope at CTIO. These datasets allow for the calibration of spectroscopic measurements against physical properties across nearly ten magnitudes of luminosity.

Results and Discussion

The authors successfully identified 12 measurable spectral lines in low-resolution spectra, including $H\delta$ and the luminosity-sensitive $Sr II$ line, that serve as objective classification criteria. Preliminary HR diagrams constructed using the new HIPPARCOS parallax data (Figure 1) demonstrate a much more complete and continuous coverage of stellar types, particularly for G- and K-type stars, compared to older catalogs.

Preliminary HR diagram constructed using HIPPARCOS parallax data, showing the distribution of stars by absolute magnitude and B-V color.

By plotting traditional luminosity classes against spectral types (Figure 2), the authors revealed clear inconsistencies in the MK system. This finding supports the transition from Roman numeral luminosity classes (e.g., III, V) to a more granular Arabic numerical system (e.g., 3.0, 3.5, 5.0), allowing for a more precise mathematical analysis of stellar populations. The results suggest that absolute magnitude acts as an additional physical parameter that the traditional 2D HR diagram fails to fully capture.

Plot of luminosity class versus spectral type, illustrating the move toward a more granular, quantitative classification system and revealing inconsistencies in traditional MK types.

Conclusion and Future Work

This research represents the initial steps toward a more rigorous stellar classification framework. While the quantitative approach shows promise, the authors acknowledge limitations, including the need for the full HIPPARCOS catalog to finalize calibrations and the challenge of defining the specific physical parameters (such as metallicity or age) that should constitute the additional dimensions of the system. Future work will focus on integrating true physical measurements to avoid the circular logic of relying on existing MK classifications for calibration.

Summary updated

Abstract

Practically all the MK spectral types known today are the result of visual inspection of photographic spectra. This includes the monumental work by Nancy Houk. Two questions related to the MK system remained unanswered all along, namely 1) How accurate are the luminosities and intrinsic colors derived from these types? And 2) Can the classification be done on the basis of measured spectroscopic criteria such as equivalent widths? A third question must also be raised, namely 3) Should we not rather classifyin a system with more than two dimensions? With the Parallax Catalogue of Also now techniques are available to produce measured spectroscopic criteria in large quantities. The first steps in this direction will be discussed here.
BibTeX
@inproceedings{stock1997towards,
  author = {J. Stock and R. Agostinho and J. A. Rose and Z. Marín and Rafael Molina},
  title = {{Towards a Multi-Dimensional Stellar Classification System}},
  booktitle = {Open},
  year = {1997},
  doi = {10.1515/astro-1997-0109},
}
Figure