Looking at the relevant wikipedia page, one can read that the graviton should be massless. Is it 100 % certain that it is massless or is there room in any "nonstandard" models for a tiny non-zero mass (which could lead to a similar surprise as the detection of the neutrino oscillation) such that the graviton (if it exists ...) could maybe selfinteract and form something like "graviballs"? (My present knowledge of GR and QFT is at the "Demystified-Level")

  • $\begingroup$ This is just a curious question... I`ll retract it if it is too annoying or stupid ;-) $\endgroup$
    – Dilaton
    Commented Jun 4, 2011 at 10:25
  • 1
    $\begingroup$ I think those are a subgroup of Spaceballs. $\endgroup$
    – Georg
    Commented Jun 4, 2011 at 10:30
  • $\begingroup$ Ha ha, maybe ;-) ... $\endgroup$
    – Dilaton
    Commented Jun 4, 2011 at 12:01

3 Answers 3


Gravitons do self-interact. That's because gravity couples to energy-momentum tensor which is non-zero even for massless particles (e.g. consider that gravity influences light). So it's certainly a possibility that they might form a bound state. But you can't see this in a linearized theory of gravity because it is a free theory (similar to electrodynamics). So you'd also have to include higher order curvature terms and quantize those, which would in principle resemble something like chromodynamics (although much harder). It is also not clear whether the theory would be consistent without including the rest of the standard model (and beyond). In any case you'd definitely need some form of quantum gravity to answer this question.

Regarding the first question: there are experimental upper bounds on masses of all massless particles. Obviously it's impossible to distinguish whether something is strictly zero or infinitesimally small. Suffice it to say that all experiments conform to the fact that the masses are zero and there is no theoretical justification whatsoever to consider other models.

  • $\begingroup$ Thanks Marek for this helpful answer. Now reading that from You I should have known that gravitons interact because they have energy. Lenny Susskind said the same thing about photons in one of his Lectures ... stupid me :-/! $\endgroup$
    – Dilaton
    Commented Jun 4, 2011 at 12:23
  • $\begingroup$ And yes, by the graviballs I meant something in analogy to the glueballs in QCD ;-) Would be interesting to know if somebody is doing that ... $\endgroup$
    – Dilaton
    Commented Jun 4, 2011 at 14:15

There exists a self-consistent, self-interacting gravitational solitons called "gravitational geon" within general relativity. As Marek said above, gravitational field is non-linear and self-interacting so energy momentum of gravity wave itself will produce bound states. You can read a bit more about it in a following Wiki page:


One important contribution that the Wiki page missed is that you can get a half-integer spin geon from pure gravity, a work by Friedman and Sorkin.


As far as I know there are not much development in quantum mechanical aspect of them.

  • $\begingroup$ Oh yes these graviational geons are interesting :-). It is just a bit a nuisance that I would have to buy the paper if the second link :-/ $\endgroup$
    – Dilaton
    Commented Jun 4, 2011 at 22:14

Two recent papers in that direction:

https://www.sciencedirect.com/science/article/pii/S0370269322001253?via%3Dihub https://journals.aps.org/prl/abstract/10.1103/PhysRevLett.127.081601

  • $\begingroup$ Hello Random number and welcome to Physics SE. Link-only answers are not very popular here because if the links go dead then the answer is instantly invalidated. Additionally, if the links do not lead to downloadable (or accessible) content then they may be useless (not the case here though). So, it is better to provide at least a brief description of the linked content and the solution it provides, so that people looking at the answer get a solution to their problem and at the same time be able to find more details in the links. $\endgroup$
    – ZaellixA
    Commented Jul 26, 2022 at 11:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.