Early breakfast to get going before 8, but the travel photographer is already gone. The valley is much narrower below the village, and the path crosses the stream a few times before settling on the left side to descend. The stream drops down and the valley becomes almost a gorge with the path occasionally running along the bottom, occasionally on one or other side. The sun hasn't reached down here yet, but from time to time there is a glimpse of sunlit rocks ahead, either a bend in the valley or, later, the wall of the Markha valley's left flank.

After two hours or so, just past another abandoned tent-restaurant, I catch up with the photographer, who is standing at the edge of the shadow of the mountain in order to get a 'sunrise' effect. I'm not sure if he sneaks in a picture of me going down the trail after I've passed, but I snuck in a picture of him taking his sunrise picture, so fair's fair.

From this point the valley starts to widen and the golden wall at its end becomes more and more prominent. It's impressively high, probably over 1000 m, with no side valleys or even chinks, just a series of steep rock chutes alternating with crags.

Finally I reach the village of Skyu at the end of the valley. As is traditional, the gonpa is locked and there is nobody about, but I manage to take advantage of the facilities, being located in Ladakhi fashion in a separate building. The village has no more than ten houses or so but there is an 'Eco Cafe' and three tea tents, one of which surprises me by actually being open. The proprietor seems to double up jobs as the local blacksmith: he has some kind of metalwork in hand on his kerosene stove, but he is happy to sell me some tea as well.

In the homestay at Shingo I'd been reading an archaeological survey of the Markha valley that I'd found on the internet. This valley is full of ruins, most of which predate historical records---but in Ladakh these only go back three hundred years or so. It can be hard to spot ruins on top of crags as they are built from the same stone as the crags, and quite often crags do a convincing impression of fortresses in themselves. The 'castle' at Skyu is a case in point: a rambling collection of walls decorating a rock buttress. If I hadn't been looking out for it I would probably not have noticed it was there.

The buildings at Skyu are spread over the alluvial fan of the stream from Shingo, and so are slightly higher than the valley floor: here the fields begin, with the path squeezed between the thorn hedge and the rocky slopes. The Markha valley is a good 500 m wide, and its flat bottom is covered with trees and cultivation. I pass under trees starting to lose their leaves with autumn and between reddening bushes. Every house seems to advertise homestay, though it seems not to have affected the general tone of life, just a little extra cash windfall from time to time.

A few km up the valley the cultivation has petered out and the scenery is more austere, with trees giving way to spiky seabuckthorn bushes. Here is another ruined fort, more compact, on a protruding rock. Further on again is the camping spot at Pentse. The guidebook says that apart from the shop run by a women's development group---now closed---there is no village here, but in fact there is, in ruins, on a shelf 20 m above the path. This is the end of the stage in the book, but it's not yet noon, so I refill my bottle and carry on.

For the next few km vegetation is limited to thin lines on one or other side of the valley, the slopes bare and the valley floor covered with braided channels divided by pebbly islands. The path sticks to one side, passes a spring marked with prayer flags, then starts to pick its way down between the river channels. An older version of the path climbs up and traverses along the cliff 20 metres up; this path had been partially supported on birch poles and rocks, but these have washed or slid away in places and not been repaired, leaving the path unusable. Here also is a large boulder with ruins perched on top, supposedly the site of the cunning despatch of a demon (or maybe king) by an old lady with an adjustable fire.

Approaching a bend in the valley a huge rock spike dominates the view. This is the end of a sharp ridge which divides the Markha valley from a tributary. There is supposedly a route along this valley (I doubt it goes across the ridge, despite the map: it's a good 500 m high) which leads to the Rabrang La, serious isolation, and ultimately Zangskar. Perhaps one for my next visit...

Round the bend, the path is hemmed in between the river and a low cliff, but this is by design: going a short way up the cliff the path then launches across the river by means of a sloping wooden bridge, the use of the cliff having effectively saved the builders almost half of the work.

I'm now going due south, but not for long: the valley bends the opposite way to regain its previous course and the path climbs up onto a low terrace decorated with chortens. Here, beside a side valley coming in from the north, is Umlung Gonpa, improbably situated 300 m up a rocky shoulder. Ruins of a fort or village are visible on the opposite shoulder, below which are sitting a European woman and her two Ladakhi guides. Evidently not intending to visit the gonpa however, as they soon move on. I discover that the trail up to the gonpa has been blocked by a large tree branch, not an insurmountable obstacle but likely to denote that the place is deserted, as apparently it normally is. I suppose I could have climbed up to see the view, but I settle for taking pictures of it from below.

The gonpa is only just visible from Umlung village, from where it decorates the rock wall blocking the view downstream. Also from here I get my first view of Kang Yatze, the mountain which watches over the whole upper Markha valley. At 6400 m it's a pyramidal hulk somewhat higher than Stok Kangri, with large snow fields on every side.

The next tiny hamlet (whose name I forget) is opposite what the archaeologists call the most important Buddhist site in the valley, a group of caves and chortens almost camouflaged on a dry terrace across the water. There is a small tea tent here and a villager stops her occupation (making mud bricks using a mould, and preventing her small son from stepping on the wet ones) to entice me to stop. She says next season they will build a path up to the place for the tourists.

Beyond here the valley continues in much the same way, the path crossing the river to visit various villages, most of which advertise homestay accommodation. There is a place called Lhatho with a (closed) tea-tent and large piles of red-stained horns. Another place advertises homestay but proves to be entirely deserted. A river crossing proves to be simple despite the reservations of the guide, though a passing horseman pauses to check I don't fall in. Another empty homestay, but finally as the light draws towards evening I find an open place. The family are mostly out harvesting, the man baling up grass and the woman carrying the bales up to the threshing(?) ground, so it's the daughter who shows me round and permits me to dry my shoes by the stove, and the cat who decides to pin me to the mattress by curling up on my lap.

Before dinner I am offered a little chang for the first time. It's good, somewhat sour, taken with a lump of tsampa powder. I hadn't planned to come all the way to Markha today, but at least that makes tomorrow's stage shorter. The family think I could get all the way to Nyimaling tomorrow at my current rate, but I think an easier day is called for.