Plain convolutional neural networks (CNNs) have been used to achieve state-of-the-art performance in various domains in the past years, including biometric authentication via eye movements. There have been many relatively recent improvements to plain CNNs, including residual networks (ResNets) and densely connected convolutional networks (DenseNets). Although these networks primarily target image processing domains, they can be easily modified to work with time series data. We employ a DenseNet architecture for end-to-end biometric authentication via eye movements. We compare our model against the most relevant prior works including the current state-of-the-art. We find that our model achieves state-of-the-art performance for all considered training conditions and data sets.