In Allen and Greenough all the examples of reflexive pronouns have them come before the verb, but Pliny the Younger in e.g. letter 6.20.11 has 'non moratus ultra proripit se effusoque cursu...' and again at 6.20.12 'paret aegre incusatque se, quod....' Is this unusual and thus an emphatic position (but it's at the end of a clause which I understood was non-emphatic), or is it colloquial, or....what? Do other authors often have this word order?
In Latin Word Order: Structured Meaning and Information, p 286, Devine and Stephens say:
In styles like that of Livy, which allow V-bar syntax, a weak pronoun can remain in the base verb phrase as a postverbal clitic, thus not raising at all; this is well attested for the reﬂexive
Multi et aliarum civitatium, qui Emporias perfugerant, dediderunt se; (Livy 34.16.5)
Nam Lacedaemonii… deviis callibus medio saltu recipiebant se; (Livy 35.30.10)
sed etiam Colophonis obsidione abscessit et Sardis recepit se; (Livy 37.31.4)
Ligures… improviso oppressi ad duodecim milia hominum dediderunt se; (Livy 40.38.1) Hasdrubal… procul ab hoste intervallo ac locis tutus tenebat se, (Livy 23.26.2)
et tum quidem ab Dio Perseus in interiora regni recepit se, (Livy 42.39.1).
Apparently, some authors, like Livy and Nepos preferentially left nouns with little individual referential value after the verb; whereas others, like Caesar virtually always placed such nouns before the verb.
For instance, Devine and Stephens say:
In Caesar's syntax you are required to say aciem instru(x)it, in Livy's syntax you prefer to say instru(x)it aciem
Authors like Livy treat such nouns as part of the event description. Authors like Caesar treat such nouns as participants in an event. The style used by the first group would allow you to leave reflexive pronouns after the verb.
The difference in style might have emerged as Latin was slowly changing from a OV language to a VO language.
If you check online e.g. a text of Cicero or Tacitus for all cases of a " se ", you will find regularly only clauses where se( .... )verbum is a bracket for an often complex attributional construction of the reflexive action, that translates into subordinate dependent clauses in modern languages.
One of the syntactic elements generating classical Latins compactness in handwriting and reducing speakers time in a republican state.
From both Plinii it is known, that they didn't write themselves by hand but used notarii. Perhaps, what you found is one of the differences between poetic authoring and dictation from the spoken language eg in teaching.