# Barebow Rose Awards

19 minutes
March 9, 2020

The Rose Award is a badge scheme run by Archery GB for the York/Hereford/Bristol rounds in the UK. To date badges have only been available for compound, recurve, and longbow archers. With the recent increase in the number of barebow archers Archery GB decided suitable scores should be added to the scheme. This article details how Mark Roberts and I decided upon suitable scores.

#### The Rose Awards

The Rose Awards are a series of badges that can be claimed for reaching certain scores on the York, Hereford, or Bristol rounds for gentlemen, ladies, and juniors respectively. The scheme goes back to (I believe) the 1844 GNAM (Grand National Archery Meeting) held in Yorkshire at which a white rose was awarded. Since then there have been a series of badges developed that move through the colours of the target from white to gold, capped off by a final purple badge. The recurve design is illustrated above, with compound archers receiving a circular badge, longbows a shield, and juniors a smaller square badge.

The scores that have to be achieved are the same for gentleman on a York, ladies on a Hereford, and juniors on a Bristol, but vary between bowstyles as follows:

Bowstyle White Black Blue Red Gold Purple
Compound 800 900 1000 1100 1200 1250
Recurve 800 900 1000 1100 1200 1250
Longbow 225 300 375 450 525 600

Table 1: Rose award scores

The scores for both recurve and compound are the same, whilst the scores for longbow are lower. As an initial reaction one might take issue with the scores for recurve and compound being the same given one typically scores higher than the other, but it should be remembered that these are also the same across age and gender, despite scores typically being different between the different rounds. Furthermore, the scores for World Archery badges are also the same across Recurve and Compound, so the Rose Award scheme is consistent with this. Of course, it would make no sense for longbow scores to be identical as the awards would be unattainable, so a separate set of scores exist.

These scores are all set out in plot 1 which also shows the Archery GB classification levels (3rd Class up to Grand Master Bowman). They have been plotted using the handicap value for each score so as to allow for comparison between different rounds (in effect using a common units system - imagine Yorks being imperial and WA rounds as being metric, handicaps allows us to view both on the same scale!).

One thing that should strike any modern target archer is the absence of scores for the barebow category. To those who have been in the sport for a while, however, this can be more readily understood. The popularity of barebow as a shooting style has increased dramatically in the last decade. Ten years ago the sight of a barebow at competitions was rare, and when they were seen they were often traditional one-piece bows. There has been an exponential growth in popularity of the style in recent years, however, accompanied by widespread adoption of the take-down barebow. This has seen the style grow to a level at which it has arguably overtaken longbow in popularity, and is sometimes even more popular than compound at competitions.

Whilst Archery GB has recognised barebow as a bowstyle for many years, it is responding to increasing demand by developing a series of scores to incorporate barebows into the Rose Award scheme (until now barebow archers could technically claim Roses, but did so as recurves). This report seeks to set out a recommendation of what appropriate scores would be to incorporate barebow archers into the Rose Awards scheme.

#### Scaling scores

Plot 1 is not particularly illuminating for our purposes of deciding on scores of a comparable standard and can be improved by re-scaling the data to a parameter $$N$$ based on the classification handicaps for each bowstyle where

$N = \frac{x - BM}{BM - GMB}$

The effect of this is to align the GMB and Bowman levels for all categories, as can be seen in the plot 2. The reason we chose bowman and GMB scores to scale the data is because they were set based upon percentiles of record status scores across the UK. As a result they are grounded in actual data. This was done some time ago, however, and since archery has increased in popularity and standards the classifications may need revisiting in the near future. This is a story for another day though, and we will simply make use of them as is here.

Based on this scaling we see that compound roses are easier to achieve, as expected, but that the standards required for the recurve and longbow roses are somewhat comparable across both bowstyle and gender. The lower longbow roses are a little more challenging than recurve, especially for gents.

This scaling does, of course, hinge on the classification systems for each bowstyle being comparable which, as previously mentioned, is an entire can of worms in itself, and something we hope to address in the future. Indeed we see that the lower barebow classifications appear anomalously lower than other bowstyles, and there is evidence to suggest that all of the barebow classification scores in particular should go up since the style has increased in popularity and standards since they were set. Nonetheless, this does give us to first-order a method by which to compare the scores we develop to the existing ones.

#### Key features of the Rose Awards

The first part of our analysis involved highlighting the main features of the Rose Awards scheme to guide our setting of scores. The key points we noted are as follows:

• The use of round scores
For recurve and compound archers the use of round scores provides a clear and definitive target to work towards. It also ensures that scores are easy to remember on the day of a shoot and do not require looking up like classifications etc. We should try to ensure that the suggested scores for barebow are set at round scores, ideally the nearest hundred or fifty.

• Regular score intervals
Each Rose award is an equal number of points from its neighbours (with the exception of compound/recurve purple). The use of equal steps means that scores are memorable, and that an archer always knows what their next award is. Furthermore, by taking regular steps between badges it ensures that they do not excessively overlap with other schemes and the classification system. We should try to ensure that the suggested scores for barebow increase in regular intervals as far as is reasonably possible.

#### Key questions

In order to guide the development process we decided to break the problem down into three smaller questions that, once answered, should assist the setting of new scores. These were as follows:

• Where is the white score?
What is the entry level to the scheme that makes an achievable first step, but presents a similar level of challenge when compared to the awards for other bowstyles?

• Where is the purple score?
What is the top level for the scheme that presents a similar level of challenge when compared to the awards for other bowstyles?

• What should the scores in between be?
Balancing the key features noted above, and compared to the awards for other bowstyles, what is the best distribution of badge scores between the top and bottom awards?

#### Data

Rather than using the aged Archery GB handicaps as validation, it would be preferable to use some recent real-world scores. To this end an invaluable resource during the project was the national longbow and barebow rankings compiled by Paul Gregory, Barry Du-Crow and Nick Hayball. This is a database of all longbow and barebow scores shot at record status competitions in the UK during 2019.

Finding similar data for compound and recurve is difficult. For the purposes of this report I scraped the results from all York/Hereford and WA 1440 rounds shot during 2019 from the ianseo website using the selenium and beautiful soup modules in python. It is suspected that there may well be biases in this data, however, since not all competitions are uploaded to ianseo, and those that are tend to be reasonably high-profile events. As such there is a danger that these scores will be biased towards the top end. In particular there are certain shoots (the Masters, University Championships, and County Team Tournament) that are biased in their entry standards towards the upper classifications. If future work is to be done in this area then efforts must be made to gather data more complete data in which we have more confidence - ideally all scores from all record status events like the LB/BB rankings.

There are a number of points for consideration when using this data that may not initially seem important. However, they could have quite a large impact on the final outcome and change our interpretations as we shall now discuss:

• Which scores should we use?

• Use all scores
This maximises the amount of data used, but has the downside that it introduces biases. The best archers tend to shoot more competitions than the average archer leading to a bimodal distribution with scores being skewed towards the top end by a small group of elite archers.

• Use an archer’s average score
This removes the biasing effect of a few top archers, but has the downside that it can be significantly affected by outliers, for example from an extremely bad weather competition. As a result taking the average is not necessarily a good representation of regular ability here.

• Use an archer’s single best score
Since Roses are awarded for an archer meeting a score threshold once, a more appropriate metric might be to use an archer’s maximum score for the year. This removes the bias due to elite archers attaining many high scores, the effects of averaging with low outliers, and is representative of the situation we are considering. This is the option I chose to use.

• Which rounds should we use?

• Use all rounds
Again, this maximises the data, but has a number of downsides as discussed below.

• Use only long-distance rounds
The LB/BB rankings contain York/Hereford, WA1440, and WA720 rounds. However, archers’ handicaps are typically higher on WA720 rounds compared to the WA1440 and York/Hereford due to the lack of a long-distance component (the analysis supporting this might appear as a future article…). There is also a discrepancy in that, for a WA720, both genders shoot the same round.

• Use only York/Hereford rounds
Expanding upon the previous point, the long distance aspect of the York/Hereford (half of the competition) makes this a particularly challenging round - as the author discovered when questing for MB scores! Indeed, the author of the handicap system David Lane noted that an archer’s handicap tends to decrease with distance. As such it does not make sense to compare York/Hereford and WA1440 results in this study.

Based on these decisions we can now plot the data we have gathered showing archers’ maximum scores on a York/Hereford round. This is done in plot 3 below as a series of violin plots for bowstyle and gender. Also included are Rose scores, classifications, and percentiles. The differing shapes of these distributions are interesting and may be discussed in a future article, but sadly there is no time here. It should also be reiterated that the recurve and compound distributions are limited, possibly biased, and would greatly benefit from increased data.

#### Answers to the key questions

We are now in the position to use this data to answer to our key questions, and propose a set of barebow Rose scores.

• Where is the white score?
Looking at plot 3 we see that for recurve and longbow the white rose lies a little below the modal score, whilst black lies around or just above. This would suggest a black barebow score of around 550-600, and a white of around 450-500 (taking scores to the nearest 50 points).
It should be noted that, whilst longbow scores are somewhat comparable between genders, barebow shows ladies typically scoring higher with the discrepancy larger still in recurve. As a result there is no ‘correct’ answer to what the ideal White Rose score should be which is why we have suggested an appropriate range of 50 points. These brackets are shown in plot 4 below.

• Where is the purple score?
Answering this question is substantially more difficult, and involves making the biggest trade off between the rounds. It is particularly complicated by the fact that there is little data around the top end of the scheme, with the purple and gold Roses rarely being awarded.
If we look at recurve scores the purple badge is virtually unattainable - indeed, there has only ever been one purple recurve Rose awarded to Naomi Folkard! - whilst for longbow award was achieved by around 1% of gentlemen and 3% of ladies. For the anomalous compound around 10-15% of archers achieve a purple badge. This is summarised in table 2 below, which shows the cumulative percentage of archers whose maximum York/Hereford score would have seen them achieve each Rose award using our 2019 dataset.
If we match to the longbow standards then this would suggest a purple barebow score around 900. If we were to look at the recurve distribution, however, then 900 would be an appropriate gold score, with the purple set somewhere around 950-1000.
Looking at the barebow rankings from 2019, the highest barebow York and Hereford scores were 947 and 945 respectively. As another measure of what an appropriate score might be, we note that the national records in barebow are 951 for gents on the York, and 1111 for ladies on the Hereford.

After some consideration we decided that 1000 is a score that remains challenging for elite women, but is still potentially accessible for gents. The 950-1000 bracket is also shown on plot 3.

Bowstyle W Bk Bl R G P
GC 100.0 97.7 90.9 63.6 29.5 15.9
LC 100.0 100.0 100.0 84.2 36.8 10.5
GR 69.8 37.8 17.0 1.9 0.0 0.0
LR 87.1 77.4 45.2 16.1 0.0 0.0
GLB 63.3 34.3 15.1 7.8 3.6 1.2
LLB 79.7 47.8 27.5 10.1 5.8 2.9
GBB-A 68.8 39.6 24.0 13.5 5.2 0.0
LBB-A 74.6 52.4 36.5 20.6 7.9 0.0
GBB-B 68.8 39.6 24.0 13.5 5.2 0.0
LBB-B 74.6 52.4 36.5 20.6 7.9 0.0

Table 2: Percentage of archers whose max. score from 2019 data achieved each Rose.

• What should the scores in between be?
Now that we have an idea about where the top and bottom of the scheme should roughly lie we can think about the intermediate scores. Bearing in mind the desire for round numbers and regularly spaced intervals between each award, we examine the cumulative percentage of barebow archers falling within 50 point intervals between 450 and 1000 on the York/Hereford round from the 2019 data. This information is shown in table 3 below. By comparing these percentages to the numbers in table 2 we can settle on a best-fit set of barebow scores that are comparable to the recurve and longbow schemes whilst also maintaining regular spacing between badges. A particular difficulty, as mentioned in the discussion of the purple Rose above, is that the data suggests that the score discrepancy between the York and Hereford rounds is perhaps most apparent in the barebow category. Based on this data we eventually decided that the 500-1000 point range, with awards spaced at 100 point intervals, offered the best balance between this and the other factors involved.

Bowstyle 450 500 550 600 650 700 750 800 850 900 950 1000
GBB 68.8 51.0 39.6 32.3 24.0 14.6 13.5 7.3 5.2 2.1 0.0 0.0
LBB 74.6 66.7 52.4 46.0 36.5 25.4 20.6 14.3 7.9 4.8 0.0 0.0

Table 3: Percentage of barebow archers from 2019 whose maximum score achieved each threshold.

Overall the following set of scores are proposed for barebow Rose awards:

White Black Blue Red Gold Purple
500 600 700 800 900 1000

Table 4: Proposed barebow Rose award scores

This makes the white Roses slightly harder on barebow when compared to other bowstyles, but provides a reasonable match across the rest of the scheme when comparing bowstyles and genders. The achievement of black - red awards are closely comparable to the longbow scheme, whilst gold and purple are more challenging as for recurve. It should also be noted that, as barebow continues to grow in popularity, scores are anticipated to increase meaning that there is a degree of future-proofing built in to the scheme.

As a footnote it should be mentioned that consideration was given to non-linear schemes so as to better match the white award, or to make each step a similar increase in ability, but it was decided that it would be best to use regularly spaced round numbers to keep new scores in line with the current scheme.

We can now return to the plots presented at the start of this article, adding on the proposed scores. Plot 5 shows the updated version of plot 2, with scores scaled by the classifications for each bowstyle.

From this it would appear that the barebow awards are similar to recurve and longbow, but the most challenging of all three. However, when this scaling was first introduced it was noted that it was based on the assumption that the classification scores were ‘correct’. We also remarked that the scores are likely out of date given they were set a while ago and archery has increased popularity and standard in the UK, especially in the barebow category.

Therefore we might instead consider scaling the scores using information from the 2019 data. To do this we take a similar approach as above, but this time use the median and interquartile range of the data as follows:

$N = \frac{x - median}{IQR}$

This has the effect of centring all the distributions about the same level, and compressing them so that the core region of the data occupies the same height. The effect of this scaling process on the score data and Rose awards is shown in plot 6 below.

This image tells a slightly different story, with the Roses now appearing more comparable to the recurve and longbow scores. As the numerical data above suggested, we see the purple scores are of a standard comparable to recurve, whilst red - white are similar to longbow. The discrepancy between the rounds is also more apparent when using this scaling, with Roses appearing to be more challenging to achieve on the York round.

As a final measure of the scores’ suitability we calculated the average number of Roses awarded annually in each category over the course of the last decade. These can be compared to the number of barebow Roses that would have been awarded (assuming archers only claimed for their highest score of the year) during 2019. Table 5 shows this data. The numerically minded may take issue with the fact that this is raw data, not presented as a percentage of the total archers or Roses awarded. This is intentional, however, as it provides a tangible reference for those who are less numerically inclined and these percentages are intrinsically contained in the preceding analysis. An archer without a mathematical or statistical background can look at these scores and see how they compare to the other categories based on their intuitive impression of typical numbers at competitions and clubs.

Bowstyle W Bk Bl R G P Total
GC 4 10 24 30 11 2 81
LC 2 3 9 13 8 2 37
GR 37 32 20 4 0 0 93
LR 19 22 14 5 0 0 60
GLB 13 11 6 4 1 1 36
LLB 7 7 4 3 1 1 23
GBB 26 23 9 5 1 0 64
LBB 22 20 9 6 3 0 60

Table 5: Average number of Roses awarded in each bowstyle over the last decade, and the number of barebow Roses that would have been awarded in 2019.

Though this report set out to propose the most reasonable scores we could for the barebow community to be included in the Rose award scheme, which we have hopefully achieved, there are a few comments and observations we have made during the course of this work.

Perhaps most important is that it is clear that there is no one-size-fits-all set of scores. Indeed, the analysis makes it clear that archers should not try too hard to compare their Roses to those held by other genders and bowstyles.

Related to this is the interesting observation that ladies on average outscore gents in all bowstyles, indicating that the 100 yard distance presents a significant challenge. This discrepancy is, perhaps, most apparent in the recurve and barebow categories.

The work also highlighted the fact that the classifications (which should ideally be directly comparable between gender and bowstyle) are due a revision - as some may have already suspected. Though barebow is perhaps the most outdated, especially in the lower classifications, the data from this report suggests that all of the bowstyles could do with a review and update. This project is currently being discussed with Archery GB, but presents a significant undertaking requiring a much more comprehensive data set of scores across multiple rounds. Watch this space…

Despite these points we have done our best to develop a set of scores that reward shooting at a similar standard of the existing schemes, whilst retaining the spirit of the Rose award scheme. We hope that these scores are well-received by the archery community, and look forward to seeing them being awarded from this summer. We also hope that they will stand the test of time as barebow continues to grow as a bowstyle and standards continue to increase.

Happy shooting.

#### Acknowledgements and references

The greatest acknowledgements must go to Mark Roberts who worked with me to come up with these score recommendations.

Arran Croggan and Katy Cumming were our point of contact at Archery GB, and coordinated the awards review focus group who also provided valuable suggestions and feedback on our work.

The work here builds on an original study by Bob Beaney and others who were the real driving force behind demanding the introduction of Rose award scores for barebow. Credit must be given to him for pursuing this and ensuring it made it into the Archery GB review.

Thanks to Paul Gregory, Barry Du-Crow, and Nick Hayball for running the national longbow and barebow rankings and sharing data for the project.

A lot of interesting information about the archery handicap system can be found in the following articles:

• The Construction of Handicap Tables for Archers, Lane, D. Letter to GNAS (1978)
• Handicap Tables, Lane, D. Toxophilus 2(1) (1979)
• The variation of accuracy with range, Lane, D. Toxophilus 2(3) (1979)
• The construction of the graduated handicap tables for target archery, Lane, D. Publicly available article (2013)
• Shooting Administritave Procedures, ArcheryGB (2016)

The data and python scripts I used to perform this analysis will appear here for downloading in due course:

• Python scripts to scrape results from ianseo
• Anonymised data used for the analysis
• Rose awards as handicaps
• Python scripts for plotting data, calculating percentages, performing analysis