Human reaction times are usually between 0.2 and 0.4 seconds. A reaction time of 0.1 is considered to be a false start (IAAF Rule 161.2). A reaction time of 0.5 is very slow. You can test reaction times here:
Reaction times consist of the time required to recognise the stimulus, and the time to take the required action. The time to process the stimulus is much less than the time to take the action (e.g. to press the button).
Reaction to an auditory stimulus is faster than to a visual stimulus. The Startle reflex is extremely fast, and is fastest in response to an audible cue behind the ear. This is a self defence mechanism.
Reaction time correlates to the strength of the signal, and to the number of signals. Reaction is fastest to a loud or bright signal; and to a combined audio and visual signal.
Although reaction times are in the tenths of a second, consistency of reaction time is usually in the low hundredths of a second. However the consistency does not have a normal distribution. A lapse of concentration causes a big delay even when other times are close together.
There are two possible methods of taking the time:
Anticipating means that the timer watches the competitor approach the line, and attempts to press the button at a time to coincide. Reacting means that the timer watches the line and reacts when the competitor is seen to cross it.
Anticipation and reaction can be measured as follows:
In timing for Athletics, reaction timing is required (England Athletics Timekeeping guide). Each timer reacts to the Start signal, and then reacts to the competitor in their allotted lane. If they were to anticipate the finish, they would produce a shorter time than the real time. If they react to the finish, then their reaction time is the same in both Start and Finish times, and cancels itself out.
In a Head race, reacting to the time means that the timer notices the boat approaching, but then turns their attention only to the line. They then react when they observe the boat to have crossed the line. Although they are aware that the boat is approaching, they need to simulate a startlement when they notice the boat to have crossed. Although the time will be later, it will be the same for all competitors and so will cancel itself out.
Unlike Athletics, it may be possible to use anticipation timing in a Head race, because there is no timing signal for the Start. As long as a timer uses the same method for all competitors, and ideally the two timers at Start and Finish use the same method, then there should be no difference.
However with a group of competitors finishing closely together it may be difficult to give the same level of anticipation to all of them. The first would receive an anticipation time, and the next a reaction time. Reaction timing is well proven in Athletics, and so it is probably safer to stick to it.
A single timer should be used at the Start, and at the Finish. This ensures that the reaction time component of the time taken is consistent between different competitors. Times from different timers should not be mixed. Reaction and anticipation should not be mixed.
If a time has to be taken from a backup timer (for example for a missing time), then the average difference in time for a number of competitor before and after the missing time should be used to find the difference in reaction times. This difference should be applied to the time inserted.
It is difficult or impossible to press a button twice within about 0.2s. This is about 1-2 meters in a Head race. As a result, one boat closely behind another may be disadvantaged.
When two crews finish within a meter or so, the timer MUST press twice (to record two finishers) and should declare a tie for time.
It would be possible to insert a time for the later boat from a second timer, but only by averaging the timing difference over a number of times. It is unlikely that this would result in a more accurate time than obtained by the first timer. It would also be possible to view a video and estimate the time gap.
Probably the best answer, knowing there is a tie for time, is to see if the time difference is significant in the overall placing of the crew. If not, then the delay can be ignored. If it is, then the video should be examined to obtain the correct time gap to 1/10th of a second and the time adjusted.
Racing regulations are very precise about the definition of the line. In a Head race the line should be distinctly marked on both sides of the course. The line may have a bias, but competitors should be aware precisely where it is and make their own judgement about favoured side.
It is very important for the timer, backup timer and video to be on the line. There is no compensation that can be made for bias of one timer compared to another, because the amount of bias depends on how close or far away the boat is. The proper way to do this is for the backup timer(s) to stand on an elevated platform behind the main timer.
The timer needs to react to seeing a boat cross the line, rather than anticipate it. A strong visual stimulus is better than a weak one. One way to help with this is to have a post that slightly obscures the timer's view of the course side of the line. The time will then be taken when the bow emerges into view again.
A large pack of boats is very difficult to time by hand. The procedure needs to be designed to handle it. There are two different types of pack:
The reason for the difficulty is that, with hand timing, there are two independent data sets: a list of boats; and a list of times. The two have to be matched up. The only way to match them up is to write something down that matches the boat to the time. For a single boat this is trivially easy. For a pack of boats there is a risk of the two data sets going out of sync. You might either miss writing down a boat number; or you might miss pressing a time. If the number of boats is different from the number of times, there is no way to know which time applies to which boat. You have to see a pack of boats to realise how hard this can be. The problem is made especially acute if a boat has lost its boat number, or the number cannot be read.
The Watch operator or Button Presser needs to anticipate in advance how many boats there will be in a stream. He might say: "We are going to get five boats close together" or "We have twelve boats and I can't see a gap. I will give you twelve". He then needs to make sure he presses that number of times.
The Writer also needs to anticipate what he is going to write down. If it is a stream of twelve then it is quite easy to write each one down as it comes up to the line. If it is a pack of five, then he needs to write down five boats numbers as best he can, and make sure there were five times. Either way, he needs to mark clearly on the list when the number of boats matches the number of times.
Most differences should be accounted for systematically (i.e. they are not randomly incorrect times). Once the systematic errors are corrected, and provided none of the boats have been missed, then the times should be fairly accurate. A systematic error would be, for example:
The clock will show times to 1/100th of a second, but we know that manual timing is not that accurate.
It is an open question whether hand times should be rounded to the nearest tenth or nearest full second. Barring error, hand timing should generaly be precise to about 0.1 seconds, i.e the timer should be able to react within about a meter of the boat crossing the line. Probably a conservative figure would be around 0.2 seconds of variation with few results outside this. However it is difficult to round times to the nearest 0.2.
International Athletics rules prescribe hand timing to the nearest 0.1 second for races inside the stadium. Races outside the stadium (marathon, triathlon) are timed to full seconds. International Cycling round hand times to the nearest 0.1 seconds.
It is also an open question how times are rounded. International Athletics round times UP. International Cycling round times DOWN. No-one rounds times to the NEAREST, as a spreadsheet would.
Assuming that the clocks themselves are accurate it would be reasonable to give hand times for a Head race with a defined timing method and a precise line rounded up to the nearest 0.1 seconds. If the timing does not follow a defined method or does not have a precise line, then times should be rounded up to the nearest second.