There is still a good agreement between theory and experiments (Table  3 ), but Eq.  19 tends to systematically overestimate the actual BNDs (by up to a factor 10 at 27.8 kPa/s).

We obtained N_STAT equal to 118 mm ⁻³ at 27.8 kPa/s, and to 25,400 mm ⁻³ at 1,000 kPa/s (Table  3 ); for comparison, the observed BNDs are equal to 80 and 5,900 mm ⁻³ at 27.8 and 1,000 kPa/s, respectively.

The comparison of theoretical and experimental densities (Table  3 ) suggests that the values calculated using Eq.  19 are slightly overestimated, though by less than one order of magnitude.

Accordingly, the BND necessary to achieve and maintain degrees of water supersaturation <Δ X_N is about 60 times as large as in the case of homogeneous nucleation (Table  3 ).

For a decompression rate of 0.1 kPa/s, the values of N_STAT are very small (Table  3 ): from 10 ⁻²  mm ⁻³ for homogeneous bubble nucleation to 1 mm ⁻³ for heterogeneous nucleation on magnetite.

$$  = 10 ³  kPa/s, and up to 10 ⁷  mm ⁻³ at 10 ⁴  kPa/s (Table  3 ).