in the south-west of Scotland,

set up a rockfall detection system

along four miles of remote hillside railway track.

The astonishing thing is:

that system is still in operation today, 140 years later.

And so far no one's been able to come up with anything better.

- The West Highland lines travel hundreds of miles

along natural mountainsides.

There are different hazards depending on where you are

on the Scottish rail network.

Through the Falls of Cruachan, the hazards there

are definitely rockfall and boulder fall.

We're now at Bridge of Orchy on the West Highland line

and the hazards here are much more natural landslides

and debris flows down the steep-sided mountains.

Rocks that are very small can pose a risk to the train operations.

So something as small as a microwave can cause a train a problem.

Something as big as a washing machine could lead to derailment

so it is a really serious issue.

Definitely stopping the hazard at source is the best option for us.

In a typical year we will spend somewhere in the region

of £40 million, and that's on things

that prevent landslides or improve the resilience

of our infrastructure to extreme weather.

- Thank you very much!

- So Scotland's railway is about 3,500 track-miles.

So we can't protect that whole length,

so we need to look at other ways of keeping the trains safe,

which is where the rockfall detection comes in.

- By modern standards, it's a really basic system.

There's no electricity.

Just these wires along the track, held under tension,

mechanically linked to the signals.

If a big boulder falls down, it hits a wire, the wire snaps,

the tension's released,

and the signal at each end of that bit of wire changes to show danger.

Repeat for 17 signals over the four miles of railway.

The system was nicknamed Anderson's Piano after its inventor

and after the humming noises

that the wires sometimes make in high winds.

- Anderson's Piano is a very old system, 140 years.

Keeping something like that going in a modern environment,

it is quite challenging.

Firstly, I guess there's the knowledge.

There's not a user manual or a downloadable book

on how to maintain Anderson's Piano.

We need to pass that down through generations

of technicians and engineers.

And then there's just getting hold of the kit.

A lot of it was bespoke, made in foundries in the Victorian era.

Some of it we've had to improvise.

Some of it is just availability of material.

So getting hold of some of the wire can be quite challenging.

And we think we've managed to find a reasonable source

that is a good enough match.

The kind of rural fence that will use that kind of wire,

the manufacturers are striving to make it stronger and stronger

and we don't need it stronger for Anderson's piano.

We need it to break when there's a rock comes through it.

It's not perfect detection in terms of screening out.

So vegetation can trip it off,

animal movements if they do get inside the boundary.

Very manual to reset.

It's a physical task that involves lifting weights

and repositioning counterbalances

and manually moving signals back into place.

So we can't do that when the trains are running,

so that's disruptive to our network.

It's a physical job that involves human beings.

There's not a big reset button that we can push.

The first we know there's a problem is when the signal goes up

and normally the first person to see that is the train driver.

We are not able to respond someone to the problem when it occurs.

We need to wait until after it's occurred,

because the first time we know about it is when a driver tells us.

- Modern rail safety rules mean that ideally,

no one should ever drive a train past a signal that's set to danger.

If that is ever needed because of a fault somewhere,

it requires checking and double checking with central control.

If a train goes past a danger signal without permission,

there's always an investigation.

If it was because the driver was negligent,

it can end their career.

But the rules for these signals have to be different,

and they go against everything that a modern train driver learns.

If these signals are at danger,

you slow the train down to walking pace,

to a speed where you will be able to stop before you hit any obstruction,

and you radio it into central control...

but you do continue.

And that would never be allowed on a new system.

- It's offering us a level of protection,

so there's a line of defence there that works quite a lot of the time.

It would be less safe without it.

Prevention is better than detection,

so stopping the rocks falling is definitely better

than detecting them when they have occurred.

So we put a great deal of effort into containing

some of the rocks on our cuttings

and into catching boulders through high-strength,

high-capacity catch fences that are anchored into the rock.

We employ all sorts of techniques like netting,

and meshing, and doweling to hold all the rock in place.

We're constantly looking for technology to improve safety on the railway.

Satellite technology that looks for ground movement from space,

and does change detection on that.

We've tried "listening fibres" through fibre-optic cables

that listen for an acoustic signature of a rock falling

or a tree falling on the railway.

The cost of that equipment is starting to come down.

We're starting to see advances in technology

which is making it more affordable to deploy.

But one of the big issues is actually power.

These things will either need a power supply,

which in remote places can be hard to come by,

or they'll run off batteries and solar panels

which need to have sufficient capacity to keep them up all year round.

Scottish winters can be quite dark.

Some of our trials have been hampered by false alerts and false alarms

and too many false alerts to run a railway.

At Bridge of Orchy we've got some detectors in place

to capture any movement.

Now, they're the wireless tilt meters that tell us if the ground's moving.

Like your spirit level on your iPhone that moves about,

they will detect a bit of movement

and we will get an alert or an alarm remotely from that.

We've then got a camera on site that we can see the site

to see whether it's a false alarm or not, so we're not needlessly stopping trains.

We've got some infrared kit on there so that we can see at nighttime.

Then infrared reflectors to see if there's been a problem

during the hours of darkness as well.

It's all remote back to base

and it only requires a technician in our control centre

to watch that and monitor that.

It's also less prone to false alarms

and we can reset it at the touch of a button.

So we know about it before the train comes,

we can capture the problem

before it manifests itself on the railway,

and we can remotely reset it and remotely monitor it.

We do an awful lot, but it's not enough to go everywhere.

And, you know, we can't physically protect every mile of railway.

It's just impracticable.

- From talking to the folks here,

I'd say that the best description of these signals

is that they are a frustrating, imperfect system

that's still better than nothing.

Back when I was researching this video, I figured

that there must be a better solution, and with all the confidence of

someone in the YouTube comments who's only watched half the video,

I thought, "Hey, couldn't they just?..."

And no. They can't, at least not yet.

[train rumbling]

How's that for timing?!