reshape2：重铸熔融数据集给出了错误的结果

Question

我感觉我遗漏了一些基本的东西，却无法弄明白。

尝试融化数据集并重新制作原始形式，但它不起作用。

library(reshape2)

df <- structure(list(measure1 = c(4.11000246380076, 4.29087381534614, 
4.30709021582289), measure2 = c(1.71250102658365, 1.78786408972756, 
1.79462092325954), measure3 = c(86.3932114427626, 84.6768654385354, 
86.9635134280892)), .Names = c("measure1", "measure2", "measure3"
), row.names = c(NA, 3L), class = "data.frame")

原始数据集：

  measure1 measure2 measure3
1 4.110002 1.712501 86.39321
2 4.290874 1.787864 84.67687
3 4.307090 1.794621 86.96351

Molten数据集：

df <- melt(df)

  variable     value
1 measure1  4.110002
2 measure1  4.290874
3 measure1  4.307090
4 measure2  1.712501
5 measure2  1.787864
6 measure2  1.794621
7 measure3 86.393211
8 measure3 84.676865
9 measure3 86.963513

尝试重铸成原始形状，但输出很奇怪：

df <- dcast(df, value ~ variable)

      value measure1 measure2 measure3
1  1.712501       NA 1.712501       NA
2  1.787864       NA 1.787864       NA
3  1.794621       NA 1.794621       NA
4  4.110002 4.110002       NA       NA
5  4.290874 4.290874       NA       NA
6  4.307090 4.307090       NA       NA
7 84.676865       NA       NA 84.67687
8 86.393211       NA       NA 86.39321
9 86.963513       NA       NA 86.96351

我错过了什么？