The Golomb-Welch conjecture (1968) states that there are no e-perfect Lee codes in Z n for n ≥ 3 and e ≥ 2. This conjecture remains open even for linear codes. A recent result of Zhang and Ge establishes the non-existence of linear e-perfect Lee codes in Z n for infinitely many dimensions n, for e = 3 and 4. In this paper we extend this result in two ways. First, using the non-existence criterion of Zhang and Ge together with a generalized version of Lucas' theorem we extend the above result for almost all e (i.e. a subset of positive integers with density 1). Namely, if e contains a digit 1 in its base-3 representation which is not in the unit place (e.g. e = 3, 4) there are no linear e-perfect Lee codes in Z n for infinitely many dimensions n. Next, based on a family of polynomials (the Q-polynomials), we present a new criterion for the non-existence of certain lattice tilings. This criterion depends on a prime p and a tile B. For p = 3 and B being a Lee ball we recover the criterion of Zhang and Ge.The Lee metric was introduced for transmission of signals over noisy channels in [13] for codes with alphabet Z p with p a prime number, then it was extended to alphabets Z q (q ∈ Z + ) and Z in [5], [6]. One of the most central question on codes in the Lee metric is regarding the 2 existence of such codes. In [6], Golomb and Welch showed that there are e-perfect Lee codes C ⊆ Z 2 for each value of e ≥ 1, and there are 1-perfect Lee codes C ⊆ Z n for each value of n ≥ 2. They also proved that for fixed dimension n ≥ 3, there is a radius e n > 0 (e n unspecified) such that there are no e-perfect Lee codes C ⊆ Z n for e ≥ e n and conjectured that it is possible to take e n = 2 (i.e. no e-perfect Lee codes in Z n exist for n ≥ 3 and e ≥ 2). This conjecture has been the main motive power behind the research in the area. A recent survey of papers on the Golomb-Welch conjecture is provided in [10]. Next we mention some of them. Explicit bounds for e n for periodic perfect Lee codes were obtained by K. A. Post [17], namely e n = n − 1 for 3 ≤ n ≤ 5 and e n = √ 2 2 n − 1 4 (3 √ 2 − 2) for n ≥ 6; and by P. Lepistö [14] who proved that an e-perfect Lee code must satisfy n ≥ (e + 2) 2 /2.1 if e ≥ 285. Recently, P. Horak and D. Kim [10] proved that the above results hold in general, that is, without the restriction of periodicity. The Golomb-Welch conjecture was also proved for dimensions 3 ≤ n ≤ 5 and radii e ≥ 2 [4], [21], [7] and for (n, e) = (6, 2) [8]. Recently, several papers have focus on the study of the Golomb-Welch conjecture for fixed radius e and large dimensions n. The case e = 2 has been treated in [11], [12] and [19]. In the linear case, it is proved that LPL(n, 2) = ∅ for infinitely many dimensions n. A new criterion for the non-existence of perfect Lee codes was presented by T. Zhang and G. Ge in [22] and it was used for the authors to obtain non-existence results for radii e = 3 and e = 4. This criterion states that if a pair (n, e) of positive integers verifies the congruence system (3) and k(n, e) is squar...