1
00:00:00,030 --> 00:00:03,689
I received an email the other day completely out of the blue from a guy called Edward

2
00:00:03,689 --> 00:00:04,080
who

3
00:00:04,080 --> 00:00:05,529
said he was in a pub quiz.

4
00:00:05,529 --> 00:00:11,579
One of the questions was how many sides does a banana have, and although he got the answer wrong the answer was five and

5
00:00:11,679 --> 00:00:14,489
So he did some more research on bananas and found that bananas

6
00:00:14,490 --> 00:00:17,760
have three parts to the inside the fruit, so his question to me, was

7
00:00:18,220 --> 00:00:21,270
how does the five on the outside line up with the three on the inside

8
00:00:22,060 --> 00:00:27,480
but that got me thinking - so this is a normal banana, it's a Cavendish banana, that's the variety -  all

9
00:00:27,730 --> 00:00:34,590
Cavendish bananas are clones of each other which means they're genetically identical. I mean a banana doesn't have seeds in it so it can't exactly

10
00:00:34,780 --> 00:00:39,840
reproduce sexually. So if you were to push right down the center of

11
00:00:40,420 --> 00:00:42,340
the banana

12
00:00:42,340 --> 00:00:44,340
you will notice that

13
00:00:44,800 --> 00:00:48,779
it splits really easily into three parts and

14
00:00:50,079 --> 00:00:52,319
that is because the flower

15
00:00:52,930 --> 00:00:59,819
The ovary of the flower had three parts to it three locules the name for like the hole in the ovary and so the fruit develops

16
00:00:59,820 --> 00:01:03,449
in three and because the banana is centered around

17
00:01:03,699 --> 00:01:04,960
the number three

18
00:01:04,960 --> 00:01:10,380
That's why it's a bit odd that you get five sides on the outside of banana because you would expect it to be a multiple

19
00:01:10,380 --> 00:01:15,500
of three, you'd expect there to be say six sides to a banana. So what do I mean by the sides of a banana?

20
00:01:15,740 --> 00:01:17,460
well if you take a banana

21
00:01:17,470 --> 00:01:21,479
preferably one that's not too ripe because it gets less noticeable the riper the banana is

22
00:01:21,549 --> 00:01:24,959
you'll see that there are ridges that run all the way from the stem

23
00:01:25,150 --> 00:01:31,889
to the flower scar and remember that bananas normally grow this way up and you can...if I was to draw on these ridges

24
00:01:32,259 --> 00:01:37,529
you'll be able to see them more clearly so this one has one, two,

25
00:01:41,259 --> 00:01:43,259
three,

26
00:01:43,329 --> 00:01:45,329
four,

27
00:01:45,909 --> 00:01:52,379
five, so this has five ridges and therefore five sides so this banana has five sides but

28
00:01:52,840 --> 00:01:58,500
Do all bananas have five sides? So like any good scientist I went and collected more data

29
00:01:58,689 --> 00:02:01,559
I actually stood by the banana section in multiple

30
00:02:01,780 --> 00:02:08,910
supermarkets going around going one two three four five, one two three...most of them had five sides and then I found one banana

31
00:02:09,129 --> 00:02:11,129
had four sides - ahh

32
00:02:11,319 --> 00:02:15,669
and that means that five sides is not universal. It only takes one

33
00:02:16,730 --> 00:02:20,379
four-sided banana to prove that not all bananas have five sides

34
00:02:20,719 --> 00:02:27,039
But I wanted to know okay well what is the average number of sides of a banana and so I needed

35
00:02:27,500 --> 00:02:31,959
yet more data I needed an even bigger sample size and

36
00:02:32,540 --> 00:02:34,840
So I went ahead and bought

37
00:02:35,269 --> 00:02:40,959
280 bananas now counting 280 bananas was not only gonna take me quite a lot of time but then

38
00:02:41,299 --> 00:02:42,920
I'd be stuck with

39
00:02:42,920 --> 00:02:46,119
280 bananas so luckily last week I was

40
00:02:46,250 --> 00:02:49,989
invited to perform at the stand up show for a Festival of the Spoken Nerd

41
00:02:50,060 --> 00:02:53,859
you probably know Matt Parker and Steve Mould, Helen Arney from YouTube and

42
00:02:54,590 --> 00:02:55,939
so

43
00:02:55,940 --> 00:03:01,320
what's a girl gonna do with 280 bananas but have a bit of an Oprah moment

44
00:03:02,620 --> 00:03:05,760
[laughter]

45
00:03:07,260 --> 00:03:09,200
If you think that's good...

46
00:03:09,200 --> 00:03:10,860
Well, get ready to catch, everyone!

47
00:03:10,860 --> 00:03:14,000
YOU get a banana!  And YOU get a banana!

48
00:03:14,000 --> 00:03:16,000
And YOU get a banana!

49
00:03:16,000 --> 00:03:18,980
And everybody gets bananas!

50
00:03:18,980 --> 00:03:20,980
And you get a banana!

51
00:03:20,980 --> 00:03:22,980
And you get a banana!

52
00:03:22,980 --> 00:03:24,860
Everybody gets bananas!

53
00:03:24,860 --> 00:03:31,700
After that mayhem I managed to get 175 bananas counted. That's a great sample size, it's really large and

54
00:03:33,169 --> 00:03:38,799
before I looked at the data in fact before I even bought that many bananas I wanted to

55
00:03:39,470 --> 00:03:44,349
hypothesize why so many bananas have 5 sides but some of them only have 4 sides

56
00:03:44,659 --> 00:03:48,608
So I looked to the biology of the banana. 
 Bananas,

57
00:03:48,979 --> 00:03:56,889
unsurprisingly, grow on banana trees although more surprisingly banana trees aren't actually trees they're herbs but moving on

58
00:03:57,109 --> 00:04:03,279
Bananas grow in hands, the whole bunch of them is a hand and you'll get about 10 to 20

59
00:04:03,680 --> 00:04:06,909
Cavendish bananas in a hand and I was looking at these pictures

60
00:04:07,819 --> 00:04:09,819
and I spotted that

61
00:04:10,009 --> 00:04:15,069
bananas always grow in two rows and with all these bananas growing together

62
00:04:15,229 --> 00:04:21,309
I looked at where the sides actually are within the hand and I noticed that

63
00:04:21,650 --> 00:04:26,320
normally you'd get one of these ridges where two bananas were touching each other and

64
00:04:26,900 --> 00:04:31,630
where there was a gap next to a banana you tend to get just one edge rather than lots of

65
00:04:32,150 --> 00:04:33,500
ridges and

66
00:04:33,500 --> 00:04:37,690
this made me wonder well maybe it's the number of neighbors that

67
00:04:38,030 --> 00:04:42,759
influences how many sides the banana has so I looked at all the scientific literature hoping to find something on

68
00:04:43,070 --> 00:04:45,999
how all the bananas squished together is there

69
00:04:46,580 --> 00:04:51,910
do they always have the same number of neighbors and I found this paper published by Kew Gardens in

70
00:04:52,520 --> 00:04:54,579
1953 called the origin of the banana

71
00:04:55,280 --> 00:05:02,289
inflorescence and what's great like I mentioned before is that all bananas are genetically identical so they all should have very very similar

72
00:05:03,259 --> 00:05:09,639
patterns of where the bananas grow within the inflorescence and I found this figure figure one

73
00:05:10,070 --> 00:05:17,679
which someone has looked at the flower buds developing and shown how they grow and just like in the pictures of the

74
00:05:18,139 --> 00:05:24,008
adult bananas you can see that it forms two rows and that these flowers form a zig-zag pattern

75
00:05:24,500 --> 00:05:25,789
there's a field of maths

76
00:05:25,789 --> 00:05:33,399
that actually looks at if you've got things growing from a point and you've got neighbors next to each other you can get a

77
00:05:33,470 --> 00:05:36,940
boundary where those two neighbors touch and it's Voronoi tessellation

78
00:05:37,340 --> 00:05:45,190
that gives you Voronoi patterns you see it all over nature and combining the Voronoi idea with this figure and

79
00:05:45,440 --> 00:05:50,769
a very handy website that had been made by a guy called Alex Beutel which allows you to draw your own

80
00:05:50,960 --> 00:05:57,970
Voronoi patterns and put the starting points in I was able to replicate figure one as if it was a Voronoi diagram and

81
00:05:58,699 --> 00:06:00,289
this one has

82
00:06:00,289 --> 00:06:05,439
16 bananas in the hand that's pretty average for a Cavendish banana and if we count them

83
00:06:06,050 --> 00:06:09,699
twelve out of the sixteen, three quarters of the bananas

84
00:06:10,370 --> 00:06:16,299
should have five sides because they've got four neighbours and then the ones on the very ends of the bunch

85
00:06:16,669 --> 00:06:20,799
have got either four or three sides and this means if my

86
00:06:21,470 --> 00:06:26,139
neighbors hypothesis my Voronoi hypothesis is correct I would expect to see

87
00:06:26,389 --> 00:06:32,019
three quarters of bananas having five sides and the other quarter having either three or

88
00:06:32,300 --> 00:06:36,340
four sides and that would give me a mean number of sides of

89
00:06:37,010 --> 00:06:41,799
4.625. Now I've got a hypothesis I can look at the data

90
00:06:42,500 --> 00:06:46,419
the 175 bananas that the audience of the comedy club counted for me I

91
00:06:47,120 --> 00:06:50,169
genuinely did not know how this was going to go

92
00:06:50,930 --> 00:06:52,930
but it looks like

93
00:06:53,480 --> 00:06:55,480
this

94
00:06:57,400 --> 00:07:01,100
I could not believe how similar these graphs looked!

95
00:07:01,940 --> 00:07:03,080
seriously, I,

96
00:07:03,080 --> 00:07:04,250
it could have been any

97
00:07:04,250 --> 00:07:07,059
distribution, it could have been normal, it could have been literally they were only

98
00:07:07,130 --> 00:07:14,439
five and then tiny number of four side if enough, but look how similar they look!  That's insane, especially because I'd not seen

99
00:07:14,720 --> 00:07:16,430
anything about

100
00:07:16,430 --> 00:07:22,600
tessellations or Voronoi patterns being used to explain the number of sides of a banana in fact I couldn't find anything explaining the number of sides

101
00:07:22,600 --> 00:07:24,600
of a banana but still

102
00:07:25,280 --> 00:07:27,020
I'd come up with this

103
00:07:27,020 --> 00:07:33,070
hypothesis and it looks so similar obviously got to do some statistics on this can't just eyeball it as much as I would want to and

104
00:07:33,320 --> 00:07:37,839
because I got a large sample size it doesn't matter that these aren't normally distributed I can still do a

105
00:07:37,910 --> 00:07:40,689
t-test on the means on the averages and

106
00:07:41,510 --> 00:07:45,279
it turns out that we get a non significant result which

107
00:07:45,919 --> 00:07:52,269
normally you'd be a bit miffed as a scientist to see a non significant result but this test is

108
00:07:52,640 --> 00:08:00,380
testing are these two means different and so non significant means it's not able to

109
00:08:00,780 --> 00:08:03,640
disprove that the means are the same

110
00:08:04,280 --> 00:08:11,470
which basically means that the means are the same which means that my two means, my predicted mean and the actual

111
00:08:11,470 --> 00:08:13,470
mean are the same

112
00:08:13,490 --> 00:08:14,120
which is

113
00:08:14,120 --> 00:08:20,739
amazing I will say that if you look at the overall distribution not just means and do either a chi-squared test which you can't really

114
00:08:20,740 --> 00:08:22,900
do because the expected values of zero or

115
00:08:22,970 --> 00:08:27,880
If you do a Fisher's exact test then you don't get the distributions coming out as exactly the same

116
00:08:28,130 --> 00:08:31,989
but given this is biology and biology is really messy I'm

117
00:08:32,510 --> 00:08:37,119
pretty impressed with this hypothesis as explained going a long way to

118
00:08:37,370 --> 00:08:44,289
explain the number of sides that a banana has and that means that the answer to how many sides does a banana have is

119
00:08:45,140 --> 00:08:46,280
actually

120
00:08:46,280 --> 00:08:48,639
4.625

121
00:08:49,160 --> 00:08:51,520
sides which Matt Parker did point out is

122
00:08:51,880 --> 00:08:55,149
five sides when rounded to the nearest whole number of sides

123
00:08:55,250 --> 00:09:01,149
But that's not the point it means you can't just say that or bananas have five sides in the same way that the average

124
00:09:01,490 --> 00:09:04,870
person has less than two legs because of the number of

125
00:09:05,540 --> 00:09:06,430
one-legged or

126
00:09:06,430 --> 00:09:06,730
no

127
00:09:06,730 --> 00:09:10,510
legged people it brings down the averages averages are allowed to have

128
00:09:10,850 --> 00:09:13,750
non integer values we're allowed to have a

129
00:09:13,910 --> 00:09:19,510
fractional number of sides of a banana and actually someone at the comedy club came up to me afterwards from the audience and

130
00:09:19,610 --> 00:09:25,659
brought me this banana which at one end has four edges and at the other end

131
00:09:25,850 --> 00:09:33,430
has five edges and therefore is a 4.625 sided banana which I thought was great I

132
00:09:33,769 --> 00:09:37,119
Hope you enjoyed this video it's been a long time since

133
00:09:37,120 --> 00:09:40,449
I've made one but I am now back you can expect to see

134
00:09:40,610 --> 00:09:47,380
many more science videos from me in your sub feed this is now my full-time job so if you or a company wants to pay

135
00:09:47,510 --> 00:09:54,369
this newly qualified Doctor of Science to make videos for you drop me a link on my email which is all on my website

136
00:09:54,680 --> 00:09:58,989
Thank you very much and see you next time which actually might not be that long