“…Facts collected in [8,3] imply that for any hyperbolic triple (k, m, l), that is, such that 1/k + 1/m + 1/l < 1, there exist infinitely many regular hypermaps of type (k, m, l) on orientable surfaces, with automorphism group isomorphic to a linear fractional group over a finite field. By the theory developed in [3] combined with the findings in this paper, to establish a nonorientable analogue of this result requires analysing conditions under which the quantity A = 4+(α+α −1 )(β+β −1 )(γ+γ −1 )−(α+α −1 ) 2 −(β+β −1 ) 2 −(γ+γ −1 ) 2 , where α, β and γ are primitive 2k-th, 2m-th, and 2l-th roots of unity in C, projects onto a non-zero square in a quotient field of the ring of algebraic integers of Q[α, β, γ] generated by a suitable prime ideal; note that for l = 2 we have γ 2 = 1 and γ + γ −1 = 0 and then A reduces to the quantity A introduced earlier. In fact, methods of Section 3 can be adapted in an obvious way to construct, for any hyperbolic triple (k, m, l) and for suitable triples (α, β, γ) of primitive roots of unity as above, an infinite number of nonorientable regular hypermaps of type (k, l, m) over linear fractional groups.…”