The two strains of "Chelsea Stage" are nearly identical, the only changes in the second being in bars 1-2 and a single note in bar 7. Although progress through the octave in the second phrase is obvious, just whether this can be resolved into a unidirectional line is not.
One possibility is shown below, in form of a "split" line where an internal ascent goes from ^1 to ^5 (beginning of the boxed notes), then a simple rising line follows to ^8. I don't find this entirely satisfying because of the sharp trajectory running toward and reaching ^9, but one can use substitution frequently found in cadences and specifically involving the dominant: ^9 substitutes for ^2 here, in the same way that, according to the traditional Schenkerian, ^7 commonly substitutes for ^2 in the descending line.
An alternative is to elevate the ^3 in the first bar of the strain, but this is decidedly less plausible. In previous posts I have observed that it is common in such small pieces as these to make an expressive "leap" above the prevailing register at the beginning of the second strain. To choose the isolated note A5 here would seriously unbalance the prevailing expressive gestures of this dance.
"Bevis Mount" is a collection unto itself -- four independent strains that bear no relation to each other beyond being complete eight-bar themes in the same key and all closing in the home key. The second strain is of interest. The close in the upper register is definite, but here again a unidirectional line seems implausible.
Finally, then, "Cave of Enchantment" is in a small ternary design with a truncated reprise and a close in the dominant for the first strain. Emphasis on ^1, ^5, and ^8 sets the frame for the first strain. The opening of the second shifts the basic idea to the dominant level, but the result is draw out the third, F#5, which is given on the first beat three times in a row before leading to G5, thus ^7-^7-^7-^8. In the reprise, then, attention is easily shifted to G5.
Thus, I would read the second strain as given below.