This article concerns the likelihoods of competing textual reconstructions in the Chester Beatty Papyrus 𝔓⁴⁶ based on the available space in its lacunose lines. To quantify these relative probabilities, the author uses statistical models of line lengths in 𝔓⁴⁶ and a recently described technique for calculating the likelihood of a reconstructed lacunose text. He first demonstrates the power and versatility of this approach with examples in Gal 4:17 and 3:1. He then revisits two more contested textual reconstructions proposed for 𝔓⁴⁶: the absence of τῷ θεῷ in Heb 11:4, suggested by G.D. Kilpatrick in 1941, and the absence of σὺν ἐπισκόποις καὶ διακόνοις in Phil 1:1, suggested by T.C. Skeat in 1995. He shows that Kilpatrick’s proposed shorter reading in 𝔓⁴⁶ is six times more likely than the longer reading in Heb 11:4, while the evidence is not decisive between the readings in Phil 1:1.