It occurred to me today that it would be possible to use some basic statistical analysis to evaluate which translations are, on the whole, more literal and which are idiosyncratic.
You might, for example, take the Latin text of Book I. Go through it and isolate all of the root nouns and verbs (for simplicity's sake), and put them in the first column of a spreadsheet under "Latin". For the second column, Perseus; whichever definition the Perseus Project suggests for that Latin word goes in column 2. Then Munro. Then a column for numerically representing the deviation from the mode; 0 for using the mode word, 1 for using an idiosyncratic word, 2 for not translating the word at all. Then Bailey and deviation, and so forth.
Then add up the deviation for each column and divide by the number of words. This value is that translator's eccentricity. A higher eccentricity for that data set suggests a less literal translator. Because Perseus cites dictionary entries including multiple translations, it will not count toward modality, nor be included in the final tally.
| Latin | Perseus | Leonard | Eccentricity | 1743 | Eccentricity | Bailey | Eccentricity | Munro | Eccentricity |
| Aeneadum | Aeneas | Rome | 0 | Rome | 0 | Aeneas | 0 | Aeneas | 0 |
| Genetrix | Mother | Mother | 0 | Mother | 0 | Mother | 0 | Mother | 0 |
| Hominum | Man | Man | 0 | Man | 0 | Man | 0 | Man | 0 |
| Divomque | God | God | 0 | God | 0 | God | 0 | God | 0 |
| Voluptas | Delight | Delight | 0 | Delight | 0 | Joy | 1 | Darling | 1 |
| Alma | Nourishing | Dear | - | Sweet | - | Life-giver | - | increase-giving | - |
| Caeli | heaven | - | 2 | heaven | 0 | heaven | 0 | heaven | 0 |
| Signa | sign | star | - | sign | - | star | - | sign | - |
| Mare | sea | main | 1 | sea | 0 | sea | 0 | sea | 0 |
Eccentricity = (X/7) where X equals the number of words for which there is a mode. Larger numbers signify more consistent outliers.
1743: 0
Bailey: 0.14
Munro: 0.14
Leonard: 0.43
This data set is obviously so small as to be meaningless, and the project is probably not worth doing with a proper set: say, the whole of book one. It could prove interesting to sample passages throughout the book, or perhaps from the beginning of each book.
...but I'm not going to do it!