claim (i) that our estimate is only based on the analysis of Raman spectra which are notoriously difficult to calculate, (ii) that we inappropriately obtained the value of f 0:75 on the basis of the Raman spectrum of a model structure with f 0:09, and (iii) that 11 B NMR spectra are insensitive to f. We here show that these claims are unsubstantiated. The accuracy of Raman intensities and frequencies as obtained within density functional schemes has amply been demonstrated for a variety of systems including molecules [3], crystals [4], and vitreous materials [5]. The vibrational mode associated to the boroxol peak is well localized. Indeed, molecules containing boroxol rings with different terminating ligands all yield a boroxol Raman peak within only 4 cm ÿ1 [6] from the boroxol frequency in v-B 2 O 3. The incoherent Raman scattering of boroxols is also supported by experiment [7]. This supports our assumption that the Raman activity of the single boroxol ring is largely independent of the value of f [1]. Our estimate of f is not only based on the Raman spectrum, but also on the interpretation of 11 B NMR chemical shifts. We showed that the isotropic chemical shift of a 11 B atom linearly depends on the average B-O-B angles of its first-neighbor O atoms [1]. When the B-O-B angle distributions for O atoms inside and outside of the boroxol rings have different mean values, two peaks appear in the NMR spectrum. For our model, the average B-O-B angles for O atoms inside and outside of boroxol rings are 119:7 0:5 and 134:4 9:2 , respectively. Using the corresponding angular distributions together with f 0:75, the simulated NMR spectrum shows the characteristic double-peak shape (Fig. 1) observed in the experiment [8]. The two peaks derive from B atoms inside and outside of boroxol rings and correspond to different average B-O-B angles: boroxol 2 120 134 =3 125 and nonboroxol 3 134 =3 134. The chemical shift separation between the two peaks agrees closely with experiment [1]. Using the same B-O-B distributions for O atoms inside and outside of boroxol rings (Fig. 1) but for f 0:09, the simulated spectrum shows a single peak with a slight shoulder on the opposite side with respect to the experiment. Furthermore, when the same angular distribution is taken for both kinds of O atoms as proposed by Swenson and Bo ¨rjesson [2], the simulated spectrum does no longer depend on f and shows a single symmetric peak, in stark contrast with the experimental result [8]. In conclusion, our work shows that NMR provides a powerful probe for f contrarily to the claims of Swenson and Bo ¨rjesson [2]. In retrospect, 11 B NMR spectroscopy provides the most direct probe for accessing the value of f through the integrated areas under the two peaks in the spectrum, without requiring the estimate of coupling factors as for the Raman spectrum.