Targets of only a few pixels are notoriously difficult to acquire. Despite many attempts at facilitating pointing, the reasons for this difficulty are poorly understood. We confirm a strong departure from Fitts' Law for small target acquisition using a mouse and investigate three potential sources of problems: motor accuracy, legibility, and quantization. We find that quantization is not a problem, but both motor and visual sizes are limiting factors. This suggests that small targets should be magnified in both motor and visual space to facilitate pointing. Since performance degrades exponentially as targets get very small, we further advocate the exploration of uniform, target-agnostic magnification strategies. We also confirm Welford's 1969 proposal that motor inaccuracy can be modeled by subtracting a ''tremor constant'' from target size. We argue for the adoption of this model, rather than Fitts' law, when reflecting on small target acquisition.