= Solution
In the observational setting of this paper, the <cosmic microwave background> favours nearly flat geometry, while cluster dynamics and large-scale structure favour a matter density near $\Omega_m\sim0.3$. The missing contribution to the <critical density> is then roughly $\Omega_X\sim0.7$. The supernova <luminosity distance> measurements require recent accelerated expansion. With present radiation negligible, $q_0<0$ requires
$$
\boxed{w< -\frac1{3\Omega_X}\simeq-0.48\quad(\Omega_X\simeq0.7).}
$$
For an X-dominated universe the threshold would instead be $w<-1/3$. Thus X cannot be ordinary pressureless matter or radiation. It must be comparatively smooth on cluster scales, so that it contributes to the background expansion without being counted as clustered dynamical matter. Its negative pressure makes its density dilute more slowly than matter, allowing late dominance. A <cosmological constant>, with $w=-1$, or sufficiently potential-dominated <quintessence> can have these properties.
The comparison with a <perfect fluid in general relativity> needs a distinction between background stress and perturbations. A homogeneous canonical scalar has <canonical scalar stress as a perfect fluid>:
$$
\rho=K+V,\qquad p=K-V,\qquad K=\dot\phi^2/2.
$$
It has isotropic pressure and no background shear stress, just as a <perfect fluid in general relativity> does. Negative pressure therefore does not, by itself, prevent a perfect-fluid description. However, an adiabatic <barotropic fluid> with the constant closure $p=w\rho$ would have $c_s^2=dp/d\rho=w$. For negative $w$, a short-wavelength density mode then has $\omega^2\simeq wk_{\rm phys}^2<0$, the <negative-w barotropic-fluid gradient instability>. Such a closure cannot give a stable, smooth propagating X component.
A canonical scalar instead has rest-frame <sound speed> squared equal to one: its spatial and temporal kinetic terms have the same coefficient. Its pressure perturbation need not obey the background relation $\delta p=w\delta\rho$. This nonadiabatic response suppresses small-scale clustering without the negative-sound-speed instability. A <cosmological constant> is a separate limiting case with no independent propagating fluid density mode. Thus the issue is the microphysics and perturbation closure, not whether the background <stress-energy tensor> has perfect-fluid form.
Back to article page
