I saw this ArXiv preprint after reading about it in the popular media.

I can not even understand the beginning of the introduction (much less the whole paper):


A Fourier transform analysis of 2.5 million spectra in the Data Release 8 of the Sloan Digital Sky Survey (SDSS) and the SEGUE 2 SDSS was carried out to detect periodic modulations contained in their frequency spectra (Trottier 2012).

where Trottier 2012 is Recherche de signaux périodiques dans des spectres astronomiques, M.Sc. Thesis, Université Laval

The modulation frequency seems to be of order 10${}^{-12}$ seconds, and the SDSS is (as far as I know) not really a particularly time-resolved data set. 1 micron wavelength light (for general reference) has a frequency of about 3x10${}^{-14}$.

I'm not asking (necessarily) if this result is right or wrong, I'd just like to understand how the SDSS can actually be analyzed to detect or even infer THz rate modulation!

  • $\begingroup$ Did you read the rest of the paper? They clearly describe their process and have several useful plots that show you that the periodic modulation frequency is not a frequency in time. $\endgroup$
    – zephyr
    Oct 25, 2016 at 18:36
  • $\begingroup$ @zephyr I can not find a clear description of the data analysis, they seem to simply refer to Trottier 2012 for all the important details, and I can't read French, so I'd hoped to get an authoritative answer from an astronomer more familliar with this. The term "modulation" in the context of technology (human or extraterrestrial) is most often (but not always) used to describe a time variation. If you can find a few sentences that clearly and unambiguously describe the analysis, that would be a good answer. $\endgroup$
    – uhoh
    Oct 25, 2016 at 22:23

1 Answer 1


The idea of Trottier & Borra is to analyze signals that might be hidden IN the stellar spectrum. I know this sounds weird; it did sound weird to me when I first heard of it too.

You know how you can decompose light in diferent colors, right? Well, we astronomers like to think of colors in terms of wavelength (with $~4500$ Angstroms being blue and $~7000$ angstroms being red) or in frequency ($1.5\times 10^6$ GHz and $2.3\times 10^6$ GHz respectively. To get these numbers, just divide the wavelengths by $c=3\times 10^8$ m/s, the speed of light). In the range of wavelengths in which the Sloan Digital Sky Survey (SDSS) works, as you can see, it is much more clear to work in wavelength, but people, for different reasons (like in this paper), can also work in frequency-space rather than in wavelength-space.

The thing is that the SDSS has spectra for millions of objects, including stars. That is, they have measurements of the flux as a function of wavelength (color) for these objects. Well, the idea in this paper is actually an idea Ermanno Borra published years ago (2012; here is the paper: http://adsabs.harvard.edu/abs/2012AJ....144..181B). In this paper, he proposes that maybe extraterrestials would like to communicate by injecting extra signals in that spectra, i.e., that they would add extra light in different colors to the stellar spectra we observe at Earth. However, he proposes that maybe they would put colors in frequency-space rather than in wavelength-space (because it is easier to modulate that with, e.g., lasers), and that these color modulations will be periodic in this frequency-space spectra. So, basically, this is what they search for in that spectra: periodic modulations in the colors, but in frequency-space rather than in wavelength. This is what they show, for example, in their Figure 4:

Plot from Trottier & Borra (2016)

The top panel has an added signal with a periodic modulation that is $10^4$ times larger than the one they detected on the real star (lower panel).

As you can see, they don't analyze a time series: they analyze the stellar flux in frequency space. Now, I think I have quite a bunch of experience with stellar spectra, and quite frankly I think this result might just be spurious signals due to instrumental problems and/or data reduction issues. I've seen similar modulations before reducing both high-resolution, mid-resolution and even low-resolution spectra (the process of going from the image to the final spectra easily introduces things like this); I wouldn't be surprised if this is the case too.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .