More on Descent Speeds

With apologies for a rather lengthy essay . . .

In 2012 a letter from me in the Fellrunner magazine led to a thread on “Descent speeds” in this forum. My interest in this topic was originally an offshoot of an attempt to find formulas relating runners’ pace to gradient (a somewhat futile project, given that pace also depends on terrain and individual runners’ characteristics, but I got it published: https://dspace.lboro.ac.uk/dspace-js..._published.pdf). My interest has now been re-ignited, and I am wondering if anyone can help with the following queries.

The data for my earlier study came from the few downhill-only races that exist, and a few up-and-down races that regularly publish split times at the summit in results lists on their websites. That doesn’t include Burnsall or the Grasmere Guides (or any other BOFRA race that I could find). However, someone has been timing runners at the summit, at least on some occasions, since we know about Ernest Dalzell’s (disputed) 2:42 descent at Burnsall in 1910, Fred Reeves’ much slower descent when he broke the record in 1977, and Reg Harrison’s 2:44 descent at Grasmere in 1957. So, two questions:

1. How was the timing at the summit done: did the summit official walk up after synchronising stopwatches, or did he stand at the summit with a pair of binoculars, starting his watch when he saw the race start? What I am really asking is, how reliable are the times stated above?

2. Are split times at the summit recorded regularly at either of these races, and if so, where could they be found? My earlier study was somewhat inconclusive about optimum gradients for vertical descent speed, but the mean gradient should almost certainly be steeper than 25 percent. Burnsall and Grasmere don’t quite satisfy this; it would be good to have summit split times from the really steep races (Alva, Coniston Gullies), but it doesn’t appear that these are ever recorded officially — or does anyone know differently?

The websites for both the (amateur) Burnsall Feast race and the BOFRA Burnsall race give excellent route information; there are differences between the two races, but the altitudes of start and finish are similar; yet the Burnsall Feast website quotes 899 feet of ascent, whereas the BOFRA website quotes 800 feet. Checking the OS 1:25000 map shows that the BOFRA estimate is certainly more accurate. This leaves Reg Harrison’s Grasmere descent as the fastest officially recorded vertical descent rate that I am aware of (1.67 m/s), but only if we accept the figure of 900 feet (274 m) of ascent. Unfortunately the Guides Race website doesn’t have detailed information on the route: can someone familiar with the route please check the OS map to confirm or modify this estimate?

If we arbitrarily decide that a “mountain” course must have at least 500m of ascent, the fastest descent rate in a “mountain" race in the UK is still Keith Anderson’s 1.365m/s at Pen-y-Fan, but the world record in the Alaskan race has been improved to 1.52m/s by Eric Strabel.

Finally, the thread two years ago also discussed optimal gradients for horizontal speed (4-minute miles, etc). The Meltham Maniac Mile was mentioned: its gradient is 7.6 percent, whereas my study of downhill data suggested an optimum a little steeper that 10 percent. But in any case, Craig Wheeler’s 3:24 at the 1993 Meltham Maniac Mile appears to still be the fastest recorded mile in history: http://holmfirthevents.co.uk/event/m...iac-road-race/ .

A welcome post because there are those who would relish more hard comparative data on race statistics such as have not been seen in The Fellrunner since Brian Martin's superb and majestic articles.

There is something fascinating about significant record times that stand for years and the older and more incredible, as in "surpassing belief", a record is the more sceptical I am. Over the years several records have been viewed with scepticism but other equivalent performances by the record holder at the time should provide a test of credibilty.

In the June 2006 Fellrunner I published a little piece about Menna Angharad's 1996 record for the Peris Horseshoe under the title "Unbeatable Women's Record?": which illustrated some prescience because it still stands. Menna's record time might have been thought incredible until one examined the records she was breaking in other long races at the time- or that she beat Sarah Rowell in this British Champ. counter into second place to earn the comment "awesome and on another level" from Sarah.

There is a line in a John Ford movie (TMWS Liberty Valance) about myth being more powerful than truth and in the case of some of the very long standing records maybe the myths have become true.

Graham, remember Crabbe. Without digging out my copy isn't there a tale about a cyclist throwing away his bidon before key climbs. Photographic evidence contradicting the story is ignored for obvious reasons, and rightly so. Never, ever let the truth get in the way of a good story. Happy New Year to you btw and to anthonykay. A truly splendid first post sir and welcome

Quote:

---

Originally Posted by anthonykay

2. Are split times at the summit recorded regularly at either of these races, and if so, where could they be found? My earlier study was somewhat inconclusive about optimum gradients for vertical descent speed, but the mean gradient should almost certainly be steeper than 25 percent. Burnsall and Grasmere don’t quite satisfy this; it would be good to have summit split times from the really steep races (Alva, Coniston Gullies), but it doesn’t appear that these are ever recorded officially — or does anyone know differently?

---

I believe split times are recorded at the BOFRA Kirk Fell ‘Wasdale Show’ race, I also think this race is very suitable for yielding modelling data for reasons I’ll go into below.

Quote:

---

A significant gap also exists in the uphill data: there are five data points from Vertical Kilometre races at gradients between 0:526 and 0:476, but the gradient of the next steepest race is 0:364. Data are also lacking at gentle gradients, both uphill and downhill. The fastest pace in the dataset is 0:127s:m��1, recorded at the Meltham Maniac Mile (m = ��0:076), while the fastest vertical speeds are 0:568m:s��1 uphill at a gradient m = 0:294 at the Grouse Grind, and ��1:474m:s��1 downhill at m = ��0:364 at the Mount Marathon Race.

---

Quote:

---

…but very uncertain due to the shortage of data at steep downhill gradients; indeed it is on steep descents that variations between individual runners tend to be most pronounced.

---

The BOFRA Kirk Fell race would probably be one of the best races to look at because the terrain is either consistently scree, or consistently grass; and it is also very steep and consistently so. These characteristics would offer a situation whereby terrain difficulties are either negligible or very consistent, giving an opportunity to achieve data for a situation where variables that might affect p(m) are either negligible or consistent when the race under scrutiny is not a road race. The record for this race is somewhere around 35mins which is staggering given the gradient and looks to indicate a course where runnability is consistent. If nothing else it would give a good point from which to look at races which are known for terrain irregularities.

Quote:

---

However, at some uphill gradient, athletes will spontaneously make the transition from running to walking. We do not have any data on the gradient at which this occurs, and there will certainly be considerable variation between athletes, but we can expect that participants in a race will choose their gait to optimise their performance.

---

The transition from running to walking and back again can often depend on what the person in front of you is doing. You may not want to refuse to drop to a walk like him/her because you plan on overtaking at a point at which your advantage can be maximised.

As a general comment on your paper i can only applaud you for being so thorough, i wasn't expecting to see differential equations. It also occurred to me that somewhere in Electrical/Electronic circuit theory there might be a model which accurately represents the one you are trying to create, since in such areas components of equations and functions could be shown to accurately represent those in your hill running model. Also this would offer opportunities to practically test scenarios using lab experiments with calibrated instruments and components.

Thank you all for your kind and welcoming comments.

With regard to long-standing records, it is notable how many current fell race records date from the 1980’s, especially Kenny Stuart’s records from 1983-5 (and his 1981 record at the BOFRA Alva race). Kenny still holds the records for both Ben Nevis and Snowdon — the Scafell Pike Race hadn’t been started while he was racing on the fells, thus depriving him of the opportunity for a complete set of British three peaks records! I am old enough to remember Kenny’s dominance of the sport in those years — and my shock when it was announced that he was going to leave the fells to concentrate on marathon running.

Turning to just one of mr brightside’s points, I agree that data from the Kirk Fell Race at Wasdale Show (which had somehow escaped my attention until now) would be really useful, for all the reasons mentioned. But I haven’t been able to find split times, or even a precise statement of the record, on either the BOFRA or Wasdale Show websites. However, on the Descent Speeds thread on this forum two years ago, Harry H Howgill gave some personal data including a 1.36m/s descent speed at this race. Does anyone know where more data from the Wasdale Show race might be found?

blackcombe time the decent from the last cp, but don't know how long theve been doing it?

Thank you Woodzy, the data on the Black Combe Race website is really good.

Hi

I don't know if this is any use to you, but this year's Skirrid fell race recorded summit times and the results can be found here.

all the best

When I started fell running it was normal for races like Three Shires (I am looking at my 1988 results) to provide CP split times (4 CP + final time) although only to "the minute" and these were obviously very marshal-reliant. Then bigger fields and the difficulty of recruiting marshals meant this practice fell away.


But now we have Sportident!

Not many hill races have splits but these days we can use strava (segments) to at least get a rough assessment of descent speeds. Here's a variety from some Scottish (and one English) races/downhill segments based on race/strava data. I've ordered them in vertical speed :

Hillend cairn to gate (part of Caerketton Downhill Race - Pentlands) :
Andrew Gilmore (Carnethy) - Time : 70s, Vertical Drop : 115m, Descent Rate : 1.643m/s, Distance : 0.4km, Pace : 4.42min/mile

Nebit Downhill Race (Ochils) :
Iain Gilmore (Carnethy) - Time : 284s, Vertical Drop : 415m, Descent Rate : 1.461m/s, Distance : 1.6km, Pace : 4.46min/mile

Jenkins Hill (Skiddaw) - Top corner (620m altitude) to bottom gate :
Adam Harris (Fife) - Time : 216s, Vertical Drop : 290m, Descent Rate : 1.343m/s, Distance : 1.1km, Pace : 5.16min/mile

The Law summit to bridge (Ochils) :
Mark Johnston (Carnethy) - Time : 345s, Vertical Drop : 450m, Descent Rate : 1.304m/s, Distance : 1.5km, Pace : 6.10min/mile

Ben Ever to Nebit col (Ochils) :
Mark Johnston (Carnethy) - Time : 225s, Vertical Drop : 260m, Descent Rate : 1.156m/s, Distance : 1.3km, Pace : 4.38min/mile

Alva Games - Summit to road :
Alasdair Dunn (Helm Hill : 2011 from strava) - Time : 332s, Vertical Drop : 370m, Descent Rate : 1.114m/s, Distance : 0.9km, Pace : 9.54min/mile

The above is a real mix of terrain and gradient. Andrew Gilmore's descent is the steepest section of the Caerketton Downhill, the split taken during his record breaking run this year (first person to beat 5mins for the course - phenomenal run). His brother Iain's stats over the whole Nebit downhill are probably equally as impressive. That Nebit race includes a relative slack gradient run in through the woods at the end so his vertical descent rate on the hill would have been higher. Adam Harris's run down Jenkins Hill path (cutting corners!) is the quickest on strava though I expect Kenny Stuart was a lot quicker! (ditto Alva Games descent?). The descent of The Law is a fast descent though very technical (rocky) & steep in bottom 100m or so which kills the speed. Alva Games similarly suffer because it is very technical and steep. I added in stats for the descent off Ben Ever in the Ochils as it is very fast on relatively un-technical terrain (grassy path) but simply isn't steep enough to get a high vertical drop rate.

So overall, I'd say the Nebit downhill is probably as close as you'd get to the ideal steepness/terrain combination for fast vertical rate of descent. Any other fast descents on strava out there?

No-one has yet mentioned the Tryfan downhill record of 680 m in 8 minutes, 1.416 m/s.

Quote:

---

Originally Posted by Woodzy

blackcombe time the decent from the last cp, but don't know how long theve been doing it?

---

There's a similar question and a bit of an answer on the black combe thread: http://forum.fellrunner.org.uk/showt...k-Combe/page31

In fact that record was broken by Pennine's Muir Morton, in front of the TV camera, but Mike claimed that they forgot to start the clock until 30 seconds into the race! Hmm?

Can any record be realistically be compared when there are so many day to day variables such as weather conditions, humidity, undefoot conditions, errosion, pitched paths etc.
IMO the Snowdon and Ben nevis Records will probably not be broken anytime soon because the path have been pitched (or as the fashionable expression goes `fixed`) and therfore a very different terrain to what they were when the records were set.

Thank you for these further nuggets of information and other comments. Some responses to some of your contributions:-

Dave_Mole: I can’t find a record time for the descent at Skirrid, but thanks for the pointer anyway.

nearlyhopeless: Caerketton is already in my dataset, but only an old record (5 mins 5 secs), so thanks for pointing out that this has been broken.
I wasn’t aware of the Nebit race, so that’s a useful new bit of data, especially since it is so steep (though not as steep as the BOFRA Alva race). I see from the SHRA website that there is also a downhill race on Bishop Hill, which I again didn’t know about.
For consistency with the rest of my data, I am only going to be using “official” timings, even if that does include a bit of flat road or field at the bottom of a descent. But these strava timings will help to check the credibility of the other data.

gej: I already have Mike Blake’s 8:00 for the Tryfan Downhill Dash in my dataset. However, the OS map shows a 917m spot height at the summit, and 306m on the road where the race finishes: 611m descent in 480s in 1.273m/s. Does anyone have any idea where the claimed 680m comes from?

JohnK: For a race with many years’ history, one can assume that the record was set in good conditions (weather and underfoot). So when comparing record times of different races, year-to-year variations of weather and underfoot conditions won’t be important; however, differences in the terrain type (e.g. tourist path vs. heather and boulders) certainly will account for much of the variation of pace between races that can’t be explained by differences of gradient.
Some of my data suggests that pitched/fixed paths may be advantageous for climbing speed, but I don’t know about their effect on descending.

Quote:

---

611m descent in 480s in 1.273m/s.

---

Anthony, the height difference is 611m from summit to car park but the distance traveled for this 611m is about 1300m at an average angle of 29 degrees according to my basic triangle maths, making the descent speed 2.7m/s. You can't use the vertical height to calculate speed, it has to be the distanced traveled ie the slope or gradient.

Slogger, when I refer to "descent speed", I mean rate of height loss (or "vertical descent rate", as I called it in my first post), not the speed down the sloping ground which you have calculated. I do think rate of height loss (and height gain, when climbing) is of interest, but you are right to say that I shouldn't call this "speed".

The problem with trying to calculate speed down (or up) a slope is that simple calculations using trigonometry or Pythagoras' theorem are only accurate if the slope is constant (although the actual error should be small in most cases, when the calculation is done this way for a real mountain with varying slope). So in my published article I never attempted to calculate true speeds over the sloping ground. I only considered vertical ascent/descent rates and horizontal pace, the latter using map distance rather than distance over the sloping ground. But if the distance for the Meltham Maniac Mile and some of the uphill "Killer mile" races in my dataset was measured using a measuring wheel on the road, then there would be a (fairly small) error in my calculated "horizontal" pace.

Thanks for the explanation Anthony, after I read your article fully I realised what you were meaning, the m/s confused me....I should have read the article first.. :-)

There's the another WFRA descent only race; the Jubilee Plunge held the first Sunday after Christmas.

Records M 10:45 Ifan Richards F 11:59 Helen Fines as stated on the website, but I think this is incorrect and the M record 10:22 Adam Peers.

Listed as 2 miles -1500ft.

My track is normally 2.2 miles 1,367ft but a compulsory route during the finishing portion made it 2.4 this year.

Helix, that's another case of a race that's already on my list, but with an old record. I have now found 10:24 for Adam Peers on the wfra website, and thanks for alerting me to possible errors in the stated distance and descent; I will need to find a map of the route.

I'll make some gpx (or KML if you prefer) routes when I've got a minute if that would help. Send me your email if you want me to make them.

The main issue for the Plunge would be that I have been led to believe that everyone used to use a different route to the one used now to descend Moel Famau itself. There is one bloke (I think it is Steve Toogood) that still uses that route and I have seen someone go in a completely different direction off the start line. I imagine the forumite Welsh Harrier (Martin) the former RO will be able tell you when the crowd opted for the new route.

Once off the mountain there are a few variations to get to the finish. I don't think anyone used to use the finish route that was imposed this year due to ice.

Thanks for the offer, helix, but it's not really worth the effort of getting supremely accurate measurements for a few races, given all the inaccuracies in my other data! Your earlier estimate of 2.2 miles agrees with my rather crude ruler-on-map estimate of distance from the 1:25000 OS map, but the 1367ft of descent seems too low: the map has a 554m spot height at the summit, and a 120m contour very close to the Golden Lion in Llangynhafal, and 434m = 1424ft.
Incidentally, although I am not familiar with either the hill or the race, I am a little surprised that there is much debate about the best route; from the map, there only appears to be one obvious quickest route (except possibly for taking the slightly longer road instead of the footpath to the finish).

That height loss (1367ft) was Strava corrected. My raw data from a barometrically corrected gps device was -1409ft.

They are finickity route options that really don't make that much of a difference. I only mentioned them as I didn't know how exacting you were trying to be in your analysis as you had mentioned different terrain types.

Incidentally I work in GIS, an have run some analysis on this route. I did this as I was assessing a dataset and wanted to look at a small area I was familiar with. According to the computer and a software package (with a licecene in excess of £1000) the old route is the better choice. I can never bring myself to do it on race day though.

It's nearly a year since I started this thread, and I have eventually got round to updating my dataset with results up to 2015, using only published results (no personal Strava data, etc.). Ian Gilmore's descent at Nebit Downfall (1.465 vertical m/s) does now appear to be the fastest vertical descent speed in a UK race, although subjectively I still find Keith Anderson's 1990 descent at Pen-y-Fan more impressive: this descent followed a 580m climb, and enabled Keith to win the race when he had only been 5th at the summit. Internationally, Eric Strabel's 1.50 m/s at Mount Marathon is still the fastest.

Overall, comparing the best fit curve to my current dataset of 91 uphill and 18 downhill records with the curve obtained in 2012, the difference is much bigger at the steep uphill end than the downhill end. This may be due to the increasing popularity of Vertical Kilometre races in continental Europe, with the world VK record going under 30 minutes last year, and new record times at several of the established VK races in recent years. When are we going to get a VK race in the UK? Ben More near Crianlarich is the ideal venue: just over 1000m climbing in 2.4km horizontally, with a very accessible start location (on the A85). Of course a VK race would be impossible in England, and the only possible route in Wales is the Watkin Path up Snowdon.