Holtite, approximately (Al,Ta,□)Al6(BO3)(Si,Sb3+,As3+)Σ3O12(O,OH,□)Σ3, is a member of the dumortierite group that has been found in pegmatite, or alluvial deposits derived from pegmatite, at three localities: Greenbushes, Western Australia; Voron'i Tundry, Kola Peninsula, Russia; and Szklary, Lower Silesia, Poland. Holtite can contain >30 wt.% Sb2O3, As2O3, Ta2O5, Nb2O5, and TiO2 (taken together), but none of these constituents is dominant at a crystallographic site, which raises the question whether this mineral is distinct from dumortierite. The crystal structures of four samples from the three localities have been refined to R1 = 0.02-0.05. The results show dominantly: Al, Ta, and vacancies at the Al(1) position; Al and vacancies at the Al(2), (3) and (4) sites; Si and vacancies at the Si positions; and Sb, As and vacancies at the Sb sites for both Sb-poor (holtite I) and Sb-rich (holtite II) specimens. Although charge-balance calculations based on our single-crystal structure refinements suggest that essentially no water is present, Fourier transform infrared spectra confirm that some OH is present in the three samples that could be measured. By analogy with dumortierite, the largest peak at 3505-3490 cm-1 is identified with OH at the O(2) and O(7) positions. The single-crystal X-ray refinements and FTIR results suggest the following general formula for holtite: Al7-[5x+y+z]/3 (Ta,Nb)x□[2x+y+z]/3BSi3-y(Sb,As)yO18-y-z(OH)z, where x is the total number of pentavalent cations, y is the total amount of Sb + As, and z ≤ y is the total amount of OH. Comparison with the electron microprobe compositions suggests the following approximate general formulae Al5.83(Ta,Nb)0.50□0.67BSi2.50(Sb,As)0.50O17.00(OH)0.50 and Al5.92(Ta,Nb)0.25□0.83BSi2.00(Sb,As)1.00O16.00(OH)1.00 for holtite I and holtite II respectively. However, the crystal structure refinements do not indicate a fundamental difference in cation ordering that might serve as a criterion for recognizing the two holtites as distinct species, and anion compositions are also not sufficiently different. Moreover, available analyses suggest the possibility of a continuum in the Si/(Sb + As) ratio between holtite I and dumortierite, and at least a partial continuum between holtite I and holtite II. We recommend that use of the terms holtite I and holtite II be discontinued.