The mixed dioctahedral-trioctahedral character of Mg-rich palygorskite has been previously described by the formula yMg5Si8O20(OH)2(OH2)4·(1–y)[xMg2Fe2·(1–x)Mg2Al2]Si8O20(OH)2(OH2)4, where y is the trioctahedral fraction of this two-chain ribbon mineral with an experimentally determined upper limit of y ≈ 0.5 and x is the FeIII content in the M2 sites of the dioctahedral component. Ideal trioctahedral (y = 1) palygorskite is elusive, although sepiolite Mg8Si12O30(OH)4(OH2)4 with a similar composition, three-chain ribbon structure and distinct XRD pattern is common. A set of 22 samples identified by XRD as palygorskite and with variable composition (0 < x < 0.7, 0 < y < 0.5) were studied to extrapolate the structure of an ideal trioctahedral (y = 1) palygorskite and to compare this structure to sepiolite. Near-infrared spectroscopy was used to study the influence of octahedral composition on the structure of the TOT ribbons, H2O in the tunnels and surface silanols of palygorskite, as well as their response to loss of zeolitic H2O. All spectroscopic evidence suggests that palygorskite consists of discrete dioctahedral and trioctahedral entities. The dioctahedral entities have variable structure determined solely by x = FeIII/(Al+FeIII) and their content is proportional to (1–y). In contrast, the trioctahedral entities have fixed octahedral composition or ribbon structure and are spectroscopically identical to sepiolite. The value of d200 in palygorskite follows the regression d200 (Å) = 6.362 + 0.129 x(1–y) + 0.305y, R2 = 0.96, σ = 0.013 Å. When extrapolated to y = 1, d200 is identical to sepiolite. Based on this analysis, we propose that palygorskite samples with non-zero trioctahedral character should be considered as members of a polysomatic series of sepiolite and (dioctahedral) palygorskite described by the new formula y′Mg8Si12O30(OH)4(OH2)4·(1–y′)[x′Mg2Fe2·(1–x′)Mg2Al2]Si8O20(OH)2(OH2)4, with 0 < x′= x < 0.7 and 0 <y′ = y/(2–y) < 0.33.