Quantisation associated with translational modes can be modelled like a particle in a box (PiB).
$$E_n = \frac{n^2h^2}{8mL^2} \qquad n \in \mathbb{Z} \, | \, n \ge 1$$
By the equation above, the zero point energy associated with a PiB is $E_1 = \frac{h^2}{8mL^2}$. However, in my statistical mechanics lectures, we set this to zero by modifying the equation as below:
$$E_n = \frac{(n^2-1)h^2}{8mL^2}$$
Atkins does this too in Physical Chemistry, claiming to do so "for reasons that will soon become clear". But, alas, I'm still not sure why this is done.
Could somebody enlighten me?