We present a detailed analysis of a large sample of spectroscopicallyconfirmed ultra-massive quiescent galaxies(${\rm{log}}(M_{\ast}/M_{\odot})\sim11.5$) at $z\gtrsim2$. This samplecomprises 15 galaxies selected in the COSMOS and UDS fields by their brightK-band magnitudes and followed up with VLT/X-shooter spectroscopy and HST/WFC3$H_{F160W}$ imaging. These observations allow us to unambiguously confirm theirredshifts ascertain their quiescent nature and stellar ages, and to reliablyassess their internal kinematics and effective radii. We find that thesegalaxies are compact, consistent with the high mass end of the mass-sizerelation for quiescent galaxies at $z=2$. Moreover, the distribution of themeasured stellar velocity dispersions of the sample is consistent with the mostmassive local early-type galaxies from the MASSIVE Survey showing thatevolution in these galaxies, is dominated by changes in size. The HST imagesreveal, as surprisingly high, that $40\ \%$ of the sample have tidal featuressuggestive of mergers and companions in close proximity, including threegalaxies experiencing ongoing major mergers. The absence of velocity dispersionevolution from $z=2$ to $0$, coupled with a doubling of the stellar mass, witha factor of four size increase and the observed disturbed stellar morphologiessupport dry minor mergers as the primary drivers of the evolution of themassive quiescent galaxies over the last 10 billion years.

We present updated constraints on the variation of the fine structureconstant, $\alpha_{\rm EM}$, and effective electron rest mass, $m_{\rm e}$,during the cosmological recombination era. These two fundamental constantsdirectly affect the ionization history at redshift $z\simeq 1100$ and thusmodify the temperature and polarisation anisotropies of the cosmic microwavebackground (CMB) measured precisely with {\it Planck }. The constraints on$\alpha_{\rm EM}$ tighten slightly due to improved {\it Planck} 2018polarisation data but otherwise remain similar to previous CMB analysis.However, a comparison with the 2015 constraints reveals a mildly discordantbehaviour for $m_{\rm e}$, which from CMB data alone is found below its localvalue. Adding baryon acoustic oscillation data brings $m_{\rm e}$ back to thefiducial value, $m_{\rm e}=(1.0078\pm0.0067) m_{\rm e,0}$, and also drives theHubble parameter to $H_0=69.1\pm 1.2$ [in units of ${\rm km \, s^{-1} \,Mpc^{-1} }$]. Further adding supernova data yields $m_{\rm e}=(1.0190\pm0.0055)m_{\rm e,0}$ with $H_0=71.24\pm0.96$. We perform several comparative analysesusing the latest cosmological recombination calculations to further understandthe various effects. Our results indicate that a single-parameter extensionallowing for a slightly increased value of $m_{\rm e}$ ($\simeq 3.5\sigma$above $m_{\rm e,0}$) could play a role in the Hubble tension.