Consider a small system of energy $\epsilon$ in contact with a heat bath of energy $E$. $\Omega_b(E)$ represents the number of accessible microstates of the bath. Then the probability of finding our system with energy $\epsilon$ is
$$P(\epsilon) \propto \Omega_b(E - \epsilon)$$
Consider the Taylor expansion of $\ln\Omega(x)$: $$\ln\Omega(E - \epsilon)\approx \ln \Omega(E)-\frac{\partial \ln\Omega(x)}{\partial x}\bigg|_{x=E}\epsilon + \frac{\partial^2 \ln\Omega(x)}{\partial x^2}\bigg|_{x=E}\frac{\epsilon^2}{2} + ...$$
The usual derivations neglect second and higher order terms. Why is this the case? I know $\epsilon$ is small but small here means small compared to $E$. I don't see how to use that to show that the higher order terms can be neglected.