Replicating a Google ngram plot in R - r

Question

I'm trying to replicate a plot from the paper Michel et al., 'Quantitative Analysis of Culture Using Millions of Digitized Books' (2011). Specifically I'm trying to make the one on the top right here:
https://pubmed.ncbi.nlm.nih.gov/21163965/#&gid=article-figures&pid=fig-3-uid-2
I know the paper used v1 of the corpus but I'm doing it with v2 as it's easier to work with. When I use the Google Ngram viewer (specifying the English 2012 corpus which corresponds to v2, a year range of 1875 to 1975, and no smoothing) I get this, which looks pretty close.
When I tried to replicate this in R/ggplot I get this:
1950 and 1883 look pretty consistent with what is happening in the viewer plot, but I can't figure out what is happening with 1910. There appears to be very few occurrences of the year '1910' in the data set in comparison to some of the other years. Would anyone with a better understanding of the Google ngrams data set be able to point me in the right direction? Should I be supplementing this with something other than just the 1-gram dataset? Does the Google ngram viewer pick out occurrences of 1-grams in a different way?
The code I've used is below. A couple of other points: 1910 and 1950 do not seem to exist as 1-grams in the v2 data set, but 1883 does. To get this to even remotely work, I had to grepl for 1950 and 1910 to get any hits (i.e. they all seem to appear as parts of date ranges like 1890-1910, or with some other characters tacked on), rather than just doing a fixed search for those years in the ngram field. I also used purrr::map_dfr to do this rather than just a dplyr::case_when in case years appeared in the same ngram picked up by a grepl (e.g. the range 1883-1910 should be a hit for both of those years, not just one).
library(ggplot2)
library(dplyr)
library(purrr)
#---- Load data ----
counts_file <- file.path("data", "total_counts.txt")
ngrams_file <- file.path("data", "google_books_1gram_eng_v2.gz")
if (!dir.exists("data")) {
dir.create("data")
}
if (!file.exists(counts_file)) {
download.file(
"http://storage.googleapis.com/books/ngrams/books/googlebooks-eng-all-totalcounts-20120701.txt",
counts_file
)
}
if (!file.exists(ngrams_file)) {
download.file(
"http://storage.googleapis.com/books/ngrams/books/googlebooks-eng-all-1gram-20120701-1.gz",
ngrams_file
)
}
one_grams <- read.delim(
gzfile(ngrams_file),
header = FALSE
)
names(one_grams) <- c("ngram", "year", "match_count", "volume_count")
one_grams_subset <- one_grams %>%
filter(year >= 1875 & year <= 1975)
total_counts_temp <- t(
read.table(
counts_file,
header = FALSE
)
)
total_counts_char <- do.call(
rbind,
strsplit(total_counts_temp, ",")
)
total_counts <- apply(total_counts_char, 2, as.numeric)
colnames(total_counts) <- c("year", "match_count", "page_count", "volume_count")
#---- Recreate plot 3A from Michel et al. (2011) ----
year_subset <- function(year_char, one_grams_data) {
one_grams_data %>%
filter(grepl(year_char, .[["ngram"]], fixed = TRUE)) %>%
group_by(year) %>%
summarise(year_count = sum(match_count, na.rm = TRUE)) %>%
mutate(year_gram = year_char)
}
plot_data <- map_dfr(c("1883", "1910", "1950"),
year_subset,
one_grams_subset) %>%
left_join(as_tibble(total_counts), by = "year") %>%
mutate(frequency = 10000 * year_count/match_count) %>%
select(year_gram, year, frequency, year_count)
ggplot(plot_data) +
geom_line(aes(x = year, y = frequency, colour = year_gram)) +
theme_minimal() +
labs(col = "ngram", x = "Year", y = "Frequency")