Fragmentation processes in eruptions are commonly contrasted as phreatomagmatic or magmatic; the latter requires only fragmentation of magma without external intervention, but often carries the connotation of disruption by bubbles of magmatic gas. Phreatomagmatic fragmentation involves vaporization and expansion of water as steam with rapid cooling and/or quenching of the magma. It is common to assess whether a pyroclast formed by magmatic or phreatomagmatic fragmentation using particle vesicularity, shape of particles, and degree of quenching. It is widely known that none of these criteria is entirely diagnostic, so deposit features are also considered; welding and/or agglomeration, particle aggregation, lithic fragment abundance, and proportion of fines. Magmatic fragmentation yields from rhyolite pumice to obsidian to basaltic achneliths or carbonatitic globules, making direct argument for magmatic fragmentation difficult, so many have taken an alternative approach. They have tested for phreatomagmatism using the fingerprints listed above, and if the fingerprint is lacking, magmatic fragmentation is considered proven. We argue that this approach is invalid, and that the criteria used are typically incorrect or incorrectly applied. Instead, we must consider the balance of probabilities based on positive evidence only, and accept that for many deposits it may not be possible with present knowledge to make a conclusive determination.