Track & Field News

[Suso]

It’s been shown that the starting system used in recent Olympic Games is defective. (See Julin, L. and Dapena, J. “Sprinters at the 1996 Atlanta Olympic Games did not hear the starter’s gun through the loudspeakers in the starting blocks”. New Studies in Athletics 18(1):23-27, 2003.) The athletes are reacting to the sound of the gun coming through the air, and not to the sound that is supposedly transmitted through an electrical wire (and therefore almost instantaneously) from a microphone on the gun to the starting block’s loudspeaker. It’s as if the loudspeakers were not working at all. (Are the loudspeakers being switched off to prevent the athletes from hearing two shots –-one through the speaker and one through the air-- and thus thinking that there has been a false start??) Whatever the reason, the athletes who are farther from the starter react to the gun later --about 0.03 seconds of additional delay with each 10-meter increment in distance.

In the men’s 4x100 final in Athens the starter was, as usual, nearest to the athlete in lane 1, and farthest from the athlete in lane 8. If the defect in the starting system has not been corrected (which it probably hasn’t), this means that the British team (lane 3) had a 0.04-second advantage at the start over the USA team (lane 5). Since the British team won only by a 0.01 second margin, with a properly working system the USA team would have won the race by 0.03 seconds.

Similarly, the starting system gave Australia (lane 1) an advantage of 0.13 seconds over Trinidad/Tobago (lane 8), but they beat Trinidad/Tobago by only 0.04 seconds. So with a properly working starting system Trinidad/Tobago would have beaten Australia by 0.09 seconds.

I mean no disrespect for the British and Australian teams. It’s just that this ongoing problem with the starting system (first noticed by statistician Lennart Julin) needs to get solved, and it probably won’t get solved unless a lot of people raise a big stink about it. (For whatever it’s worth, on the gymnastics fiasco I’m on the side of the Korean and Russian gymnasts.)

[STL_Runner]

Our pathetic excuses have reached a new low. WE LOST!! Get over it.

[Suso]

STL, relax. I'm not an American.

[STL_Runner]

>STL, relax. I'm not an American.

Sorry, this was about the tenth post I've seen on various sites with new excuses about why we were robbed. I lost it.

[trackstar]

That crappy 2nd handoff robbed us more than any starting system.

[Run DMC]

1st pass wasn't that great either. There is no sign in the replay of a slow start out of the blocks. There really is no excuse for 1,3,4 in the 100m final NOT to win the relay. Just lack of baton practice.

On a bright note, Britain did not drape a flag around Mo Greene. Even better, the US team didn't make excuses.

[STL_Runner]

>1st pass wasn't that great either. There is no sign in the replay of a slow
>start out of the blocks. There really is no excuse for 1,3,4 in the 100m final
>NOT to win the relay. Just lack of baton practice.

On a bright note,
>Britain did not drape a flag around Mo Greene. Even better, the US team didn't
>make excuses.

Crawford's reaction time was horribly slow, so slow that you have to believe maybe the speaker behind his blocks wasn't working. I don't know what the official reaction time was, but it was slower than .3. I think Gatlin left a little late, because Crawford had to slow down just to make sure he'd pass the baton to Gatlin within the exchange zone. Coby Miller, who shouldn't have been on the relay in the first place, left way too early, ran in the middle of the lane (instead of on the inside where he's supposed to be) and messed around with the baton when he should have been accelerating. If you want to adjust the baton in your hand, you can always just move your fingers around. I've done it several times and it doesn't take much skill.

[mike renfro]

The US lost because Miller had his lunch eaten by both the Nigerian and Brit 3rd legs. Mo could make up one, not both. Shit happens.

[Sprinter_1997]

You are absolutely correct. I watched and was dismayed; the 1st pass was also crap. Not one excellent pass in three. Just one great pass wins this by a stride. I am sick. You may have read Greene's comment to the press (from the Official Athens website): this team has perhaps two or three practices since the three-country meet.

[JRM]

>Whatever the reason, the athletes who are
>farther from the starter react to the gun later --about 0.03 seconds of
>additional delay with each 10-meter increment in distance.

The reasoning is that the speed of sound is roughly 350m/s in dry air. Thus, assuming that the blocks in each lane are about 10m apart, one should expect a delay of 10/350 ~= 0.03s in arrival time of the sound waves. This isn't new research; it's basic physics.

I sincerely doubt the sound was not being transmitted through the speakers. Furthermore, for every lane add on 0.03s. So, those in the extreme outer lanes were robbed of even more time.

Interesting academic speculation, but not a reason for the loss.

[Suso]

JRM:

>The reasoning is that the speed of sound is roughly
>350m/s in dry air. Thus, assuming that the blocks in each lane are about 10m
>apart, one should expect a delay of 10/350 ~= 0.03s in arrival time of the
>sound waves. This isn't new research; it's basic physics.

The blocks in each successive lane are actually 7.67 m apart. The basic physics is that sound travels through the air at 350 m/s. What is NEW RESEARCH is the finding that athletes start their motions with systematically larger and larger delays with increasing distance from the starter's gun, when we should expect the starting times not to be related at all to distance from the starter's gun because the starting system should carry the sound of the gun simultaneously to every speaker. Also, the delays increase with distance by the same amounts that one would expect if the athletes were reacting to sound traveling through the air and not through the speakers. Quite a coincidence, isn't it?

>I sincerely doubt
>the sound was not being transmitted through the speakers.

I understand your disbelief, and I would share your disbelief if I had not seen the data myself. It's uncanny how closely the systematic delays in the starting times match what one would predict from the speed of sound through the air. (See data below.)

> Furthermore, for
>every lane add on 0.03s. So, those in the extreme outer lanes were robbed of
>even more time.

That is correct, although it is roughly 0.02 seconds per lane that are added. You have to consider that each successive lane is 7.67 m --not 10 m-- further ahead along the curve. (If we want to be ultra-accurate, we would need also to keep in mind that the athletes' blocks are staggered around a curve, and thus each athlete is not quite 7.67 m farther away from the gun than the athlete in the previous lane. Also, the fact that the starter usually stands on top of a podium has a tiny effect on the progression of distances to the various blocks. But these are small factors. The "big picture" is that, in each successive lane, it takes approximately an additional 0.02 s for sound to reach the athlete through the air.)

Here are data from the Julin & Dapena paper. There are 4 columns. The first column shows the lane numbers. The 2nd column shows the median reaction times for each lane in the men's 4x100 at the Goeteborg World Championships, where a good-working starting system ("silent gun") was used. You can see some random fluctuation from one lane to the next, but no trend. The 3rd column shows the corresponding data for the Atlanta Olympic Games. You can see a clear increasing progression from one lane to the next. The 4th column shows how we would expect starting times to progress, assuming the same "real" starting time as the Goeteborg average (column 2), plus the time it took the sound of the gun to travel from the gun to each lane in Atlanta (based on pure physics). You can see that it matches pretty closely the Atlanta starting times (column 3). The match is even clearer if you actually plot the data on a graph. It's too bad I can't post a graph here.

1 0.133 0.176 0.189
2 0.150 0.212 0.202
3 0.132 0.204 0.221
4 0.151 0.238 0.242
5 0.147 0.269 0.262
6 0.150 0.300 0.281
7 0.151 0.276 0.300
8 0.142 0.331 0.317


>Interesting academic speculation, but not a reason for the
>loss.

No speculation here, except for the doubt whether the problem has been fixed by now --which I bet it has not.

If the sound of the gun reached you 0.04 s after it reached the other guy, and you lost the race by 0.01 s, I'd say that the delay produced by the defective equipment made you lose the race. Would anyone argue against this? Sure, with better baton passes the Americans probably would have won in spite of the defective equipment. The Americans rightfully paid a price for their bad passes, but that price should have been to beat the British team by only 0.03 s, not actually to lose the race by 0.01 s. The defective equipment (if indeed it has not been fixed yet) is what increased their time enough to drop them to 2nd place.

[expat]

Um, now shall we take into account the tightness of the bends?

[Suso]

>Um, now shall we take into account the tightness of the bends?

Expat:

Good point, and ideally it would be good to make such a correction, but nobody really knows quantitatively what is the effect of the tightness of the bends. I think there are some data out there, but they are not trustworthy enough. But in any case the IAAF's policy is not to make corrections of any sort in marks. That is probably why there is nothing on the rulebook about taking into account the tightness of the bends.

However, the IAAF rulebook does say that the sound of the gun should reach every athlete at the same time:

IAAF Rule 129-3: ... Where loudspeakers are not used in races with a staggered start, the Starter shall so place himself that the distance between him and each of the athletes is approximately the same.

(Rule 129-3 is longer, but that is the relevant part.) Well, loudspeakers were used in Athens, but they probably did not work (the defects of this type of starting system have been known for some time), so in effect this is the same as not using the loudspeakers.

I am going along with the IAAF general policy of not adjusting race times. I am not proposing that corrections be made in the times of 4x100 races, but that the starting system should function properly in the future, so that no correction is needed. The 0.04-second delay that I have dealt with is simply a way to show what would have happened if the starting system had worked properly. It tells us that the starting system did cost the USA the gold medal. If IAAF Rule 129-3 had been followed, the USA would have won.

[Daisy]

..................A.................B.................C
>1..........0.133..........0.176..........0.189
>2..........0.150..........0.212..........0.202
>3..........0.132..........0.204..........0.221
>4..........0.151..........0.238..........0.242
>5..........0.147..........0.269..........0.262
>6..........0.150..........0.300..........0.281
>7..........0.151..........0.276..........0.300
>8..........0.142..........0.331..........0.317

A reaction times with speaker working and silent gun
B reaction times from Atlanta 4x100 (averages)
C predicted reaction times if speakers not working

>No speculation here, except for the doubt whether the problem has been fixed by now --which I bet it has not.

This data for Atlanta looks quite convincing to me, however, I was a bit suspicious if this was actually occurring in Athens.

I checked the reaction times in the men’s two semi finals as well as the final. They are as follows for each lane.

lane.....semi#1........semi#2.........final ----> semi time
1..........0.152..........0.162..........0.199 ----> 0.199
2..........0.211..........0.212..........0.159 ----> 0.138
3..........0.156..........0.146..........0.232 ----> 0.162
4..........0.149..........0.206..........0.266 ----> 0.211
5..........0.237..........0.153..........0.300 ----> 0.289
6..........0.158..........0.289..........0.295 ----> 0.152
7..........0.199..........0.406..........0.372 ----> 0.149
8..........0.252..........0.138..........0.344 ----> 0.156

From Suso’s analysis above the two semi finals look fine (conform with the A column) but the final does look suspiciously slow on the outside lanes (more like the C column).

The fourth column is to confirm that the slower RT’s in the final are not due to some athletes having slow reaction times as their norm. The fourth column represents the RT for the athlete in the final from his respective semi final (not necessarily in the same lane).

In summary, for the final the inside lane reaction times seem comparable, however, the outside lanes reaction times are significantly slower, when compared to the semi final times. For example, lane 2 was 0.159 in final and the same athletes RT was 0.138 in the semi#2 (lane 8). Conversely, lane 7 was 0.372 in final and the same athletes RT was 0.149 in the semi#2 (lane 4).

Clearly this is not a statistically relevant analysis since there are so few data points and the circumstances of each race are different (athletes physiology with regard to RT, crowd noise, nervousness etc.) This cursory analysis does, however, give credibility to Suso’s hypothesis that the speakers were not working for the final.

[Suso]

Daisy:

Wow, where did you get those reaction times? The IAAF site that I have been using (http://www.iaaf.org/oly04/results/byevent.html) showed the reaction times of all individual sprint events, but not for the relays.

[Daisy]

>Wow, where did you get those reaction times?

http://www.athens2004.com/en/resultsAthletics/results

Look at results and they all have reaction times.

[Suso]

Does anyone know if there is a way to post a graphic (jpg) here?

[Rob]

..

[Suso]

Looking at data from a single race makes it very difficult to see any underlying trends, because there are physiological differences among the individual athletes that mask the trends. To notice trends, you need to look at "average" data for each lane. However, the true "average" often is not a good way to look at this, because a single lopsided value can mess up the whole average. A different measure, called the "median", is a better way to avoid the effects of one or two crazy values.

The data from Goeteborg and from Atlanta were all "medians". This means that, if you had for instance 7 reaction times (from 7 races) in a certain lane, you would put them in order from the smallest time to the largest time, and then pick the 4th time (the one in the middle). Most of the values from Goeteborg and Atlanta were based on 7 or 8 races. (I guess they had quarter finals in those meets.)

Here in Athens, we only have data for 3 races. Therefore, there is bound to be more random "noise" left in the data, due to random physiological differences between the individual athletes taking the start in the various lanes. But still, the median values of the Athens men's 4x100 races (the medians of the values given by Daisy), show the same rising trend as the Atlanta data and what would be expected from athletes reacting to a gun sound that traveled through the air and not electrically to the loudspeakers --see data below. There is just more noise in the Athens data because we only have three races to work with.

Again, if you actually plot the data on a piece of paper the presence of a trend becomes much clearer.

..................A.................B.................C.................D
>1..........0.133..........0.176..........0.189..........0.162
>2..........0.150..........0.212..........0.202..........0.211
>3..........0.132..........0.204..........0.221..........0.156
>4..........0.151..........0.238..........0.242..........0.206
>5..........0.147..........0.269..........0.262..........0.237
>6..........0.150..........0.300..........0.281..........0.289
>7..........0.151..........0.276..........0.300..........0.372
>8..........0.142..........0.331..........0.317..........0.252

A reaction times with speaker working and silent gun (medians)
B reaction times from Atlanta 4x100 (medians)
C predicted reaction times if speakers not working
D median reaction time (SF and final) in Athens

[Daisy]

>Most pathetic excuse in the long history of excuses.

I'm not making excuses. I'm from the UK, but let's face it, if the speakers don't work it could be a problem in the 4X100 and 200m. It depends where the gun is located with respect to all the runners. I'm not sure if it matters so much for the 400m since the starts are often 'lazier'.

[Daisy]

>Here in Athens, we only have data for 3 races. Therefore, there is bound to be more
>random "noise" left in the data, due to random physiological differences
>between the individual athletes taking the start in the various lanes. But
>still, the median values of the Athens men's 4x100 races (the medians of the
>values given by Daisy), show the same rising trend as the Atlanta data and what
>would be expected from athletes reacting to a gun sound that traveled through
>the air and not electrically to the loudspeakers.

We differ a bit in our interpretation here since I think it is likely that the speaker do work some of the time but on other occassions they do not work. So looking at medians could dilute out your trends.

Clearly the only fair way to do this is have a "silent gun". Then, if someone forgets to hook up the speakers, there will be no sound when the gun is fired.

[Suso]

The problem occurs in all races. It is just most noticeable in the 4x100 because the distances from the starter's gun vary by so much. The error in the 400 m and 400 mH races is of exactly the same size as in the 4x100. If the differences in the finishing times are very large, the problem does not produce changes in the athletes' places. It is only when the final times are very close that the athletes' places are affected. But the times of the athletes are always affected. Everyone is losing a few hundredths --some a larger number of hundredths than others.

[JRM]

>Again, if you actually plot the data on a piece of paper the presence
>of a trend becomes much clearer.
>

I have yet to see the paper, but when you say a trend becomes clear, does that include error bars? That is, how significant is the trend? Can you give be some statistical figures to support it? I'd be interested to know.

[Suso]

>I have yet to see the paper, but when you say
>a trend becomes clear, does that include error bars? That is, how significant
>is the trend? Can you give be some statistical figures to support it? I'd be
>interested to know.

JRM:

Check the graph at: http://home.insightbb.com/~j.dapena/starts/atlanta-goeteborg-men.jpg
The black dots indicate the median reaction time for each Goeteborg lane. (The good "silent gun" method was used in Goeteborg.) The reaction times look like what we should expect: all lanes have similar reaction times (avg = 0.145 s). The red dots indicate the median reaction time for each Atlanta lane ("loud" gun). Obvious problems: All lanes have longer median reaction times than Goeteborg, and the larger the lane number, the larger the median reaction time. Also, the red dots fit very well with what one would have expected if the athletes had heard the gunshot through the air (blue curve). The close fit between the red dots and the blue curve is the "clear trend" that I was referring to. A straight line fitted mathematically to the blue curve shows an increment of 0.019 seconds per lane, while a straight line fitted to the red dots shows an increment of 0.021 seconds per lane, very similar. (It looks like a duck, it walks like a duck ...)

If the values used had been means (averages), it would have been possible to show error bars ("standard deviations") around each mean value. However, in order to minimize the effects of any lopsided values, the values used were medians. This was a good idea on Lennart Julin's part, because most of these data points are based on only 7 or 8 races, and even a single crazy starting time from someone who fell asleep would really distort the results if averages (means) had been used. The problem for what you (JRM) would like to see is that, because the median was used, it's not possible to calculate and plot error bars. I don't have available the data for each individual Atlanta race, so I could not make a graph with means and error bars even if I wanted to, sorry.

With respect to Athens, check now the graphs at: http://home.insightbb.com/~j.dapena/starts/athens-men.jpg
The top three graphs show the three men's races. The one on the left (SF1) shows a very slight uphill trend, the middle one (SF2) a noisier pattern but again mainly uphill from left to right, and the graph on the right (F) a clear-cut uphill trend. You could interpret this (as Daisy proposes) as a sign that in some races (SF1 and SF2) the starting system worked OK, and that there was a problem only in the Final (F). However, this is unlikely, because the average of the Goeteborg reaction time medians for all lanes (silent gun) was 0.145 s, but in races SF1 and SF2 from Athens all but one (out of 16) values were larger than 0.145. This suggests that it took the athletes more time to perceive the gun sound in Athens' SF1 and SF2 races than in Goeteborg. Also, all three graphs have upward slopes. Straight lines fitted mathematically to the dots of SF1, SF2 and F show average increments of 0.009 seconds per lane, 0.014 seconds per lane, and 0.025 seconds per lane, respectively. So two of them show smaller increments than the 0.019 seconds per lane that we would expect based on the speed of sound, and the other one larger (although all three show an uphill trend, when the use of loudspeakers should produce a flat trend as in Goeteborg). What is happening is that with such small samples (one race apiece!) there is a lot of room for random effects. Whenever the athletes in the outside lanes have better physiological reaction times than the athletes in the inside lanes, this tends to reduce the slope of the graph, and vice versa. If we use the medians of the three races (shown by the red circles in the bottom graph), this tends to reduce the random effects. Indeed, straight lines fitted mathematically to the red dots show increments of 0.020 seconds per lane, essentially the same as the 0.019 seconds per lane that we would expect based on the speed of sound.

I also looked at the Athens women: http://home.insightbb.com/~j.dapena/starts/athens-women.jpg
The two semifinals (SF1 and SF2) again showed upward trends, but the medians of the three races (red dots in the bottom graph) showed an average slope (0.011 seconds per lane) that was only slightly larger than half of what was expected (0.019 seconds per lane). Maybe this was due in part to the fact that lane 7 of the final (F, the right graph at the top) had a very low start time value. Another likely factor is that a total of three races is probably not enough to calculate precise values for the slope of the trend from lane to lane. It would be much better to have data from about 7 or 8 races, as was the case for Atlanta or Goeteborg. Still, keep in mind that the positive trends for the women also support the idea of non-functioning loudspeakers. If the starting system were working properly, the trend should be flat, not uphill.

[Track fan]

To quote Dennis Mitchell" We didn't lose. They won...."

[Daisy]

>The one on the left (SF1) shows a very slight uphill trend, the middle one (SF2) a noisier pattern but again mainly uphill from
>left to right, and the graph on the right (F) a clear-cut uphill trend. You could interpret this (as Daisy proposes) as a sign that
>in some races (SF1 and SF2) the starting system worked OK, and that there was a problem only in the Final (F). However, this
>is unlikely, because the average of the Goeteborg reaction time medians for all lanes (silent gun) was 0.145 s, but in races SF1
>and SF2 from Athens all but one (out of 16) values were larger than 0.145.

Carefull you don't over interpret the semi final times.

Crawfords slow (lane6) start may well have been due to apathy. He knew they would get through easily so didn't feel the need to get a lightening start. The only real outlier in SF2 is Ghana (lane7) with a 0.406 and they never qualified. That could easily be explained by a bad start as opposed to delayed start. The main reason for believing this is that lane 8 in SF2 got a good start of 0.138. The same runner got a similar time in the final 0.159. So it is hard to believe the speakers were not functioning in SF2.

[tafnut]

Isn't this rather irrelevant? Mo was talking about pushing the WR. Had they 'only' run 37.60-37.80, this would not have been a problem. The USA 4x1's need to assume they are going to lose EVERY time, so they will be properly motivated to practice and then execute. Instead, just the opposite is usually true.

[Daisy]

>Isn't this rather irrelevant?

I think it matters little with regard to the 4X100m. I'm from the UK so I'm happy to see the Brits gain a hundreth of a second, or so.

But what is the point of having speakers if there is a chance they won't work? Forget the details of this thread and look at the take home message.

It is fairer that the starter uses a silent gun, especially if we insist on timing to 100ths and seperating athletes to 1000th's of a second on the photo finish. Given our infatuation with timing accuracy it seems assinine that we accept that one athlete on the stagger may hear the gun 4 hundreths of a second later than another. Would you feel the same way if the US was in lane 7 and the Brits were in lane 2 and the speakers malfunctioned?

[Daisy]

>Maybe this was due in part to the fact that lane 7 of the final (F, the right graph at the top) had a very low start time value.

Looking at the womens data (4X100) that Suso compiled there is one person who is reacting significantly faster than the other women. In the semi final and the final she had reaction times of 0.155 and 0.137 respectively. These were the reaction times for Katleen de Caluwe (Belgium), is she known for good starts?

[Daisy]

>The problem for what you (JRM) would like to see is that, because the median was used, it's not possible to calculate and
>plot error bars. I don't have available the data for each individual Atlanta race, so I could not make a graph with means
>and error bars even if I wanted to, sorry.

Also it's pretty tough to put meaningful error bars on these things because the athletes and conditions are always changing and there are not enough races to analyse. I suggest JRM not look at this as a scientific experiment but rather a casual observation that needs to be looked at in more detail.

Suso's hypothesis, that the speakers don't always work, while not proven, should not be discounted out of hand. His initial evidence is quite compelling.

[Dutra]

>This subject topic just has to be the most pathetic excuse in a long history of
>US relay excuses.>>

Sherlock. Read the entire thread.

[Realist]

Yeah, forget the U.S. part of the equation. If as system as flawed as this one appears to be is in use SOMEBODY is getting robbed in each race.

[Suso]

More evidence that the speakers are not working: Take a look at the men's 4x100 starting times from last year's Paris World Championships ("silent gun" meet = good starting system),

http://www.iaaf.org/WCH03/results/gender=M/discipline=4X1/combCode=hash/roundCode=f/index.html

... and compare them with the men's Athens starting times (loud gun = defective starting system),

http://www.athens2004.com/en/AthleticsMen/results?rsc=ATM401101&frag=ATM401101_C73E

OK, here they are, distilled for you:

..................A.................B
>1..........0.132..........0.199
>2..........0.160..........0.159
>3..........0.184..........0.232
>4..........0.148..........0.266
>5..........0.158..........0.300
>6..........0.150..........0.295
>7..........0.132..........0.327
>8..........0.160..........0.344

A reaction times men's 4x100 final Paris 2003 (silent gun)
B reaction times men's 4x100 final Athens 2004 (loud gun)

This time, don't worry about the trends. Just look at the difference in the SIZES of the reaction times in the two meets.

And, while this is the 4x100 race that showed the largest contrast between Athens and Paris, it was not an exceptional occurrence. In fact, all the Athens 4x100 races had slower starting times than the two Paris races for which there are available data (the men's and women's finals):

The median starting time in the Paris 4x100 final (silent gun) was 0.155 for the men, and 0.156 for the women. (No other starting times were reported by the organizers, as far as I know.) (And I'd like to point out that the median of the medians from all the men's Goeteborg 4x100 races --another silent gun meet-- was 0.149.)

In contrast, here are the median starting times for the six 4x100 races from Athens (loud gun):

men SF1 = 0.179
men SF2 = 0.184
men F = 0.281

women SF1 = 0.255
women SF2 = 0.250
women F = 0.263

(I'd like to point out also that the median of the medians from all the men's Atlanta 4x100 races --another loud gun meet-- was 0.254.)

So every single available 4x100 race from World Championships (Goeteborg, Paris) shows better starting times than than every single available 4x100 race from the Olympic Games (Atlanta, Athens). Can it possibly be that sprinters are always wonderful starters at the World Championships (Goeteborg, Paris), but turn into pathetic starters every time that the Olympic Games come around? That makes no sense. This is just another manifestation of the fact that the speakers did not work in Athens, like they did not work in Atlanta. (It's too bad that no start times were made available for the 4x100 in the Sidney Olympics, another meet that used the loud gun.)

It's not that the speakers fail occasionally. When the loud gun is used (as is the case in the Olympic Games), there can be no question that the speakers failed in 5 out of the 7 races available --see the graphic in ...

http://home.insightbb.com/~j.dapena/starts/summary.jpg

[And there are indications that something may not have been quite right in the remaining 2 races (the men's Athens semifinals) because their starting times are somewhet elevated and because they both show an increasing trend in starting time with lane number as I showed in a previous posting, but let's leave it at 5 bad ones out of 7.]

I find it very hard to believe that engineers would be that incompetent during at least two different Olympic Games. If I believed that, I would never get on an airplane or drive my car. My best guess is that the speakers are turned off on purpose to prevent the athletes in the farthest lanes from confusing the two sounds that they would hear (the one carried through the wire and the one that travels through the air) with a false start. I don't know if this is the reason for the observed problem, it's just the only plausible explanation that I can come up with. If this is indeed what is happening, it implies that, to avoid one problem (the misperception of a false start), a new problem is being created (non-use of the very system designed to prevent farther athletes from being penalized by the extra time that it takes the sound to reach them through the air). And we are left thinking (mistakenly) that the problem of the slow speed of transmission of the sound signal through the air has been solved, when it has not.

A couple of years ago, I made a rough experiment at a meet that used the loud gun system. I positioned myself just outside the track so that one of the athletes in a 400-meter race was starting at a distance of about 2 or 3 meters in front of me and toward my right, while the starter was at a much larger distance in front of me and toward my left. I was trying to distinguish the direction from where the sound of the gun came. I heard loud and clear the sound of the gun that came from my left, but nothing coming from the starting block. Maybe it was just too faint for me to notice it ...

On a somewhat separate issue ...

Daisy: When trying to figure out if there is a problem in the starting system or not, I prefer not to make speculations on whether this athlete might have been apathetic and that other athlete might have had a bad start. This can be dangerous, because human nature makes one tend to find good reasons to throw away the data points that do not agree with one's own hypothesis, and to keep the ones that do agree. I have been including all the data that I can get my hands on, throwing nothing out. The inescapable conclusion is that, when using the loud gun system, there is a problem that occurs most of the time, and possibly all the time.

[Suso]

Another possibility is that the speakers are always "working" in the loud gun systems, but the microphone is overwhelmed by the intensity of the gun sound, and thus the speakers deliver a sound that does not increase instantaneously to full intensity. The sound might come out through the speakers as shhhhhhhhh-BANG! The "BANG" part may take so long that the sound traveling through the air arrives before the speaker's "BANG". Are there any sound systems experts out there to confirm or deny the possibility of this?

I was once told a (possibly apocriphal) story about Ed Moses. When starting blocks first came into use, it was noticed that Ed Moses, in addition to running like a demon, also had much better starting times than anyone else. When asked about it, he answered that the sound he heard through the speakers sounded to him like shhhhhhhhh-BANG! The other runners started with the BANG; he started with the "shhhhh". I don't know if the story is true, but there it is as an alternative explanation for why the speakers don't seem to work properly with the loud gun system.

[Daisy]

> shhhhhhhhh-BANG!

Interesting. I've never heard that before. So even with a silent gun the athletes would have to be savvy enough to know which sound to start on.