diff --git a/doc/demo-journal/images/journal-benchmark.jpg b/doc/demo-journal/images/journal-benchmark.jpg
new file mode 100644
index 00000000..5b5b0309
Binary files /dev/null and b/doc/demo-journal/images/journal-benchmark.jpg differ
diff --git a/doc/demo-journal/slides-journal.md b/doc/demo-journal/slides-journal.md
index 7731c9a7..9d115c3b 100644
--- a/doc/demo-journal/slides-journal.md
+++ b/doc/demo-journal/slides-journal.md
@@ -227,7 +227,7 @@ echo 1 | sudo tee /sys/devices/system/cpu/intel_pstate/no_turbo
 cabal run bench-sqlite
 
 for i in $(seq 10); do
-  for j in $(seq 6 12); do
+  for j in $(seq 2 12); do
     cabal run bench-journal -- $((2**$j)) >> /tmp/bench-journal-$j.txt
     cabal run bench-sqlite  -- $((2**$j)) >> /tmp/bench-sqlite-$j.txt
   done
@@ -236,7 +236,7 @@ done
 
 # Benchmark statistics
 
-* TODO: show graph
+![](./images/journal-benchmark.jpg)
 
 # Amdahl's law vs the Universal scalability law
 
diff --git a/doc/demo-journal/slides-journal.pdf b/doc/demo-journal/slides-journal.pdf
index 50dc54ad..d8f7cee6 100644
Binary files a/doc/demo-journal/slides-journal.pdf and b/doc/demo-journal/slides-journal.pdf differ
diff --git a/src/sut/dumblog/tools/benchmark-plot.r b/src/sut/dumblog/tools/benchmark-plot.r
index 8ec99663..6328d798 100755
--- a/src/sut/dumblog/tools/benchmark-plot.r
+++ b/src/sut/dumblog/tools/benchmark-plot.r
@@ -1,59 +1,400 @@
 #!/usr/bin/env Rscript
 
-df <- read.csv("/tmp/bench-stats.csv")
-
-bartlett.test(df$throughput ~ df$workload, data=df)
-
-#message("If p-value >= 0.05, use var.equal=TRUE below")
-#
-#t.test(df$throughput ~ df$workload, data=df,
-#       var.equal=TRUE,
-#       conf.level=0.95)
-
-#df$workload <- factor(df$workload , levels=c(
-#  'journal-6', 'sqlite-6',
-#  'journal-7', 'sqlite-7',
-#  'journal-8', 'sqlite-8',
-#  'journal-9', 'sqlite-9',
-#  'journal-10', 'sqlite-10',
-#  'journal-11', 'sqlite-11',
-#  'journal-12', 'sqlite-12',
-#  'journal-13', 'sqlite-13',
-#  'journal-14', 'sqlite-14'))
-
-df$workload <- factor(df$workload , levels=c(
-  'journal-5000',  'sqlite-5000',
-  'journal-6000',  'sqlite-6000',
-  'journal-7000',  'sqlite-7000',
-  'journal-8000',  'sqlite-8000',
-  'journal-9000',  'sqlite-9000',
-  'journal-10000', 'sqlite-10000',
-  'journal-11000', 'sqlite-11000',
-  'journal-12000', 'sqlite-12000',
-  'journal-13000', 'sqlite-13000',
-  'journal-14000', 'sqlite-14000',
-  'journal-15000', 'sqlite-15000'))
-
-#df$workload <-
-#  factor(df$workload , levels=c('journal-6',  'journal-7', 'journal-8',
-#                                'journal-9', 'journal-10', 'journal-11',
-#                                'journal-12', 'journal-13', 'journal-14',
-#                                'sqlite-6',  'sqlite-7', 'sqlite-8',
-#                                'sqlite-9', 'sqlite-10', 'sqlite-11',
-#                                'sqlite-12', 'sqlite-13', 'sqlite-14'))
-#
-#colours = c(rep("red", 9), rep("blue", 9))
-
-X11()
-
-# https://stackoverflow.com/questions/37694234/two-boxplots-on-the-same-graph
-
-boxplot(df$throughput ~ df$workload,
-        #names=c(2^6, 2^7, 2^8, 2^9, 2^10, 2^11, 2^12, 2^13, 2^14,
-                #2^6, 2^7, 2^8, 2^9, 2^10, 2^11, 2^12, 2^13, 2^14),
-        xlab="# of clients",
-        ylab="throughput",
-        )#col=colours)
-
-message("Press return to continue")
-invisible(readLines("stdin", n=1))
+Input4=("
+workload  throughput
+journal-4                  7024.48
+journal-4                  7144.94
+journal-4                  7187.51
+journal-4                  7061.21
+journal-4                  6826.70
+journal-4                  6683.50
+journal-4                  6896.12
+journal-4                  6919.83
+journal-4                  7307.94
+journal-4                  7102.50
+sqlite-4                  9017.16
+sqlite-4                  8809.81
+sqlite-4                  8613.81
+sqlite-4                  8467.63
+sqlite-4                  8600.22
+sqlite-4                  8628.48
+sqlite-4                  8244.84
+sqlite-4                  8075.54
+sqlite-4                  8875.07
+sqlite-4                  8792.31
+")
+
+df4 = read.table(textConnection(Input4),header=TRUE)
+bartlett.test(df4$throughput ~ df4$workload, data=df4)
+
+message("If p-value >= 0.05, use var.equal=TRUE below")
+
+t.test(df4$throughput ~ df4$workload, data=df4,
+       var.equal=TRUE,
+       conf.level=0.95)
+
+Input8=("
+workload throughput
+journal-8                  6765.83
+journal-8                  6977.90
+journal-8                  7089.01
+journal-8                  6481.04
+journal-8                  6717.82
+journal-8                  6426.51
+journal-8                  6837.21
+journal-8                  6454.03
+journal-8                  6910.78
+journal-8                  6712.71
+sqlite-8                 7561.52
+sqlite-8                 7811.76
+sqlite-8                 7904.59
+sqlite-8                 7773.24
+sqlite-8                 7411.04
+sqlite-8                 7631.94
+sqlite-8                 7771.31
+sqlite-8                 7641.79
+sqlite-8                 8232.62
+sqlite-8                 7501.84
+")
+
+df8 = read.table(textConnection(Input8),header=TRUE)
+bartlett.test(df8$throughput ~ df8$workload, data=df8)
+
+t.test(df8$throughput ~ df8$workload, data=df8,
+       var.equal=TRUE,
+       conf.level=0.95)
+
+Input16=("
+workload throughput
+journal-16                  6661.66
+journal-16                  6759.78
+journal-16                  7032.06
+journal-16                  6662.63
+journal-16                  6803.18
+journal-16                  6579.79
+journal-16                  6492.99
+journal-16                  6724.89
+journal-16                  6595.02
+journal-16                  6643.49
+sqlite-16                 6165.21
+sqlite-16                 6070.99
+sqlite-16                 6421.87
+sqlite-16                 6536.94
+sqlite-16                 6540.05
+sqlite-16                 5714.42
+sqlite-16                 6162.20
+sqlite-16                 6454.41
+sqlite-16                 6141.35
+sqlite-16                 6010.05
+")
+
+df16 = read.table(textConnection(Input16),header=TRUE)
+bartlett.test(df16$throughput ~ df16$workload, data=df16)
+
+t.test(df16$throughput ~ df16$workload, data=df16,
+       var.equal=TRUE,
+       conf.level=0.95)
+
+Input32=("
+workload throughput
+journal-32                  6745.54
+journal-32                  6435.81
+journal-32                  6325.01
+journal-32                  6511.75
+journal-32                  6379.79
+journal-32                  6293.02
+journal-32                  6595.44
+journal-32                  6453.33
+journal-32                  6371.21
+journal-32                  6298.35
+sqlite-32                  6060.26
+sqlite-32                  5860.26
+sqlite-32                  6105.28
+sqlite-32                  6047.47
+sqlite-32                  6067.61
+sqlite-32                  5815.05
+sqlite-32                  6016.67
+sqlite-32                  6152.74
+sqlite-32                  6077.45
+sqlite-32                  5954.26
+")
+
+df32 = read.table(textConnection(Input32),header=TRUE)
+bartlett.test(df32$throughput ~ df32$workload, data=df32)
+
+message("If p-value >= 0.05, use var.equal=TRUE below")
+
+t.test(df32$throughput ~ df32$workload, data=df32,
+       var.equal=TRUE,
+       conf.level=0.95)
+
+Input64=("
+workload throughput
+journal-64   6193.51
+journal-64   6492.17
+journal-64   6093.18
+journal-64   6283.59
+journal-64   6253.92
+journal-64   6363.34
+journal-64   6131.53
+journal-64   6468.23
+journal-64   6147.20
+journal-64   5957.34
+sqlite-64    6358.80
+sqlite-64    5939.43
+sqlite-64    6324.36
+sqlite-64    5882.36
+sqlite-64    5941.00
+sqlite-64    6141.81
+sqlite-64    5792.27
+sqlite-64    5987.80
+sqlite-64    6202.03
+sqlite-64    5980.05
+")
+
+df64 = read.table(textConnection(Input64),header=TRUE)
+bartlett.test(df64$throughput ~ df64$workload, data=df64)
+
+message("If p-value >= 0.05, use var.equal=TRUE below")
+
+t.test(df64$throughput ~ df64$workload, data=df64,
+       var.equal=TRUE,
+       conf.level=0.95)
+
+Input128=("
+workload throughput
+journal-128                6140.07
+journal-128                6021.51
+journal-128                6167.16
+journal-128                6446.74
+journal-128                6134.16
+journal-128                6053.93
+journal-128                6236.71
+journal-128                6354.00
+journal-128                6014.61
+journal-128                5914.42
+journal-128                  6275.96
+journal-128                  6130.14
+journal-128                  6069.00
+journal-128                  6226.67
+journal-128                  6242.63
+journal-128                  5969.20
+journal-128                  6094.10
+journal-128                  6111.73
+journal-128                  5845.98
+journal-128                  6143.18
+sqlite-128               7258.15
+sqlite-128               6995.89
+sqlite-128               6825.82
+sqlite-128               7054.20
+sqlite-128               6895.98
+sqlite-128               6849.55
+sqlite-128               6704.36
+sqlite-128               6704.68
+sqlite-128               6849.77
+sqlite-128               6847.92
+sqlite-128                  6886.70
+sqlite-128                  6873.38
+sqlite-128                  6567.82
+sqlite-128                  6879.67
+sqlite-128                  6786.69
+sqlite-128                  6807.89
+sqlite-128                  6940.66
+sqlite-128                  6526.45
+sqlite-128                  6527.77
+sqlite-128                  6842.96
+")
+
+df128 = read.table(textConnection(Input128),header=TRUE)
+bartlett.test(df128$throughput ~ df128$workload, data=df128)
+
+t.test(df128$throughput ~ df128$workload, data=df128,
+       var.equal=TRUE,
+       conf.level=0.95)
+
+Input256=("
+workload throughput
+journal-256                6174.16
+journal-256                6009.00
+journal-256                6133.48
+journal-256                6118.66
+journal-256                6251.49
+journal-256                6444.50
+journal-256                6289.08
+journal-256                6090.86
+journal-256                6107.82
+journal-256                6178.65
+sqlite-256               5738.44
+sqlite-256               5733.58
+sqlite-256               6605.20
+sqlite-256               5822.83
+sqlite-256               5990.19
+sqlite-256               5467.91
+sqlite-256               6216.10
+sqlite-256               6368.61
+sqlite-256               6097.44
+sqlite-256               6095.78
+")
+
+df256 = read.table(textConnection(Input256),header=TRUE)
+bartlett.test(df256$throughput ~ df256$workload, data=df256)
+
+t.test(df256$throughput ~ df256$workload, data=df256,
+       var.equal=TRUE,
+       conf.level=0.95)
+
+Input512=("
+workload throughput
+journal-512                5889.80
+journal-512                5967.63
+journal-512                5871.33
+journal-512                6129.90
+journal-512                6038.78
+journal-512                6000.86
+journal-512                5780.60
+journal-512                5756.72
+journal-512                6042.59
+journal-512                5895.52
+sqlite-512               4851.46
+sqlite-512               4175.24
+sqlite-512               5001.94
+sqlite-512               6861.95
+sqlite-512               4624.16
+sqlite-512               6735.40
+sqlite-512               6603.62
+sqlite-512               6625.08
+sqlite-512               6482.68
+sqlite-512               6430.65
+")
+
+df512 = read.table(textConnection(Input512),header=TRUE)
+bartlett.test(df512$throughput ~ df512$workload, data=df512)
+
+t.test(df512$throughput ~ df512$workload, data=df512,
+       var.equal=TRUE,
+       conf.level=0.95)
+
+Input1024=("
+workload throughput
+journal-1024                 6060.03
+journal-1024                 5713.12
+journal-1024                 5984.37
+journal-1024                 5642.23
+journal-1024                 5733.45
+journal-1024                 5664.29
+journal-1024                 5667.71
+journal-1024                 6007.26
+journal-1024                 5949.81
+journal-1024                 5859.07
+sqlite-1024                4411.89
+sqlite-1024                4082.66
+sqlite-1024                4422.64
+sqlite-1024                3827.05
+sqlite-1024                4253.10
+sqlite-1024                5100.49
+sqlite-1024                3986.22
+sqlite-1024                4175.26
+sqlite-1024                4204.46
+sqlite-1024                4017.86
+")
+
+df1024 = read.table(textConnection(Input1024),header=TRUE)
+bartlett.test(df1024$throughput ~ df1024$workload, data=df1024)
+
+t.test(df1024$throughput ~ df1024$workload, data=df1024,
+       var.equal=TRUE,
+       conf.level=0.95)
+
+Input2048=("
+workload throughput
+journal-2048                 5541.96
+journal-2048                 5411.09
+journal-2048                 5571.17
+journal-2048                 5606.22
+journal-2048                 5276.86
+journal-2048                 5259.30
+journal-2048                 5467.38
+journal-2048                 5464.42
+journal-2048                 5355.09
+journal-2048                 5496.44
+sqlite-2048                3504.50
+sqlite-2048                3840.96
+sqlite-2048                3737.51
+sqlite-2048                3874.28
+sqlite-2048                3541.43
+sqlite-2048                4334.83
+sqlite-2048                3773.90
+sqlite-2048                3805.36
+sqlite-2048                3843.81
+sqlite-2048                3773.41
+")
+
+df2048 = read.table(textConnection(Input2048),header=TRUE)
+bartlett.test(df2048$throughput ~ df2048$workload, data=df2048)
+
+t.test(df2048$throughput ~ df2048$workload, data=df2048,
+       var.equal=TRUE,
+       conf.level=0.95)
+
+Input4096=("
+workload throughput
+journal-4096                 4797.12
+journal-4096                 5047.38
+journal-4096                 4978.69
+journal-4096                 4781.63
+journal-4096                 5006.89
+journal-4096                 4916.87
+journal-4096                 4821.37
+journal-4096                 4983.61
+journal-4096                 5004.28
+journal-4096                 4905.44
+sqlite-4096                3593.53
+sqlite-4096                3377.08
+sqlite-4096                3545.17
+sqlite-4096                3429.96
+sqlite-4096                3763.35
+sqlite-4096                3509.91
+sqlite-4096                3499.49
+sqlite-4096                3393.35
+sqlite-4096                3665.23
+sqlite-4096                3718.05
+")
+
+df4096 = read.table(textConnection(Input4096),header=TRUE)
+bartlett.test(df4096$throughput ~ df4096$workload, data=df4096)
+
+t.test(df4096$throughput ~ df4096$workload, data=df4096,
+       var.equal=TRUE,
+       conf.level=0.95)
+
+c4    <- c(7015.473, 8612.487)
+c8    <- c(6737.284, 7724.165)
+c16   <- c(6695.549, 6221.749)
+c32   <- c(6440.925, 6015.705)
+c64   <- c(6238.401, 6054.991)
+c128  <- c(6129.595, 6831.315)
+c256  <- c(6179.770, 6013.608)
+c512  <- c(5937.373, 5839.218)
+c1024 <- c(5828.134, 4248.163)
+c2048 <- c(5444.993, 3802.999)
+c4096 <- c(4924.328, 3549.512)
+
+df <- data.frame(c4, c8, c16, c32, c64, c128, c256, c512, c1024, c2048, c4096)
+names(df) <- c("4", "8", "16", "32", "64", "128", "256", "512", "1024", "2048", "4096")
+
+colours=(c("red", "blue"))
+
+# X11()
+jpeg("/tmp/journal-benchmark.jpg",
+     width=640)
+
+barplot(height=as.matrix(df),
+        xlab="# of concurrent clients",
+        ylab="Throughput (reqs/s)",
+        beside=TRUE,
+        col=colours)
+legend ("topright", c("journal", "sqlite"), fill=colours)
+
+#message("Press return to continue")
+#invisible(readLines("stdin", n=1))
+dev.off()