Sampling Strategies and Data Model

 A corner of Book Row (Fourth Ave. at Astor Place), c. 1935.
From the NYPL Digital Gallery, Image ID: 708165F.

One of the things I’ve been struggling with is how to narrow down the sample of bookstores I’m going to map. In looking to the past, I have decided to use Book Row as my starting point. Although I know that bookselling took place in NYC prior to the 1890s/early 1900s, the concentration of bookstores once located there is a significant historical and spatial benchmark, so it makes sense to begin there. However, since I’m also going to map some of the new crop of indie bookstores that have sprouted up in the last 5-10 years, that still leaves me with an approximately 100-year time span to cover.

I could limit my sample by neighborhood (i.e. just Book Row and the East Village), but that would exclude some of the wonderful examples I’ve found so far in Brooklyn, Harlem, and elsewhere in Manhattan. Since one of my arguments is centered on indie bookstores’ role as vital cultural centers and community spaces, I’m seeking out those that fill that capacity. Which stores function as something beyond a place for customers to purchase a product and leave? Which bookstores offer (or once offered) opportunity for personal interaction, learning, and creative connections, and where are they located?

Another issue for me is the breadth of information available on some of these stores. Certain ones, such as the Strand and Gotham Book Mart, were open for decades. How do I go about sampling each of their individual histories? So far, I’ve been looking to find artifacts or information on which authors and significant cultural events connected to the stores. The National African Memorial Bookstore and the way in which it served as a platform for Malcolm X. was an exciting find. As much as I think it would be useful to get a lot of points plotted on URT, it might be a better strategy for me to focus on a smaller number of bookstores and do a more in-depth exploration of each rather than many at the surface level.

Here is a really rough, tentative outline of my data model:


•Historic Bookstores
•Contemporary Bookstores


•Date Opened
•Date Closed (if applicable)
•Associated Authors
•Type of Books:  New, Overstock, Used, Rare, Antiquarian
•Community functions: workshops, performances, readings, etc.
•Images: exterior, interior, events
•Audio or Video: recorded readings, oral histories, performances

I have historic and contemporary bookstores as different entities because I think it would be useful to visually separate the new crop of stores from the old, most of which no longer exist. However, what would I do with a store like the Strand which is certainly historic but also contemporary since it’s still in operation? Might “Closed Bookstores” and “Open Bookstores” be a better categorization? Some of attributes could function as entities too so that I could make reference links between them, but I’m not sure which ones would be worth pulling out. Some bookstores had multiple locations (and unfortunately it’s harder to find their addresses at a certain date than at others!) while others did not. Some changed ownership while others did not. There is still so much to nail down. If anyone has comments or suggestions, I’d love to hear them!


  1. These are interesting challenges, Farah. First, I agree with you that, “as much as…it would be useful to get a lot of points plotted on URT, it might be a better strategy…to focus on a smaller number of bookstores and do a more in-depth exploration of each rather than many at the surface level.”

    As you know, we’ll be prototyping our projects this week, and we’ll have time for individual conferences next week — so hopefully between this week and next you can find answers to some of your questions.

    I agree that Book Row is a perfect place to start. You could potentially jump right from the 19th c. to today — but are there perhaps bookstores that opened sometime in-btw the late 19th century and today that are worth mapping?

    I think your confusion over your entities might be in part an issue of semantics. Separating out “historic” from “contemporary” stores implies that historic stores can’t also be contemporary — which isn’t true for the historied stores that are still in existence, and still vibrant. What if, instead, you labeled your entities “recently opened” and “whatever-the-opposite-of-recently-opened-is”? Of course that wording is terribly awkward, but perhaps by finding the right terminology you can free yourself from entity names that either introduce contradiction or create methodological challenges.

    1. Thanks for your comment, Shannon. I like the idea of “recently opened” and “opposite-of-recent.” I’ll work on trying to figure out better terms for these entities in the next week or two.

  2. Farah, if you have not already done so, you should definitely read the book “Book Row: An Anecdotal and Pictorial History of the Antiquarian Book Trade” by Marvin Mondlin and Roy Meador (Carroll & Graf, 2005). Lots of information about long-gone Fourth Ave. stores, including street addresses, opening and closing dates, etc.

    1. Thanks for your comment, Sebastian. Yes, I checked that book out of the library and it has become one of my main sources of information for this project. It contains a lot more data than I’ll be able to plot but is a fascinating read nonetheless.

Leave a Reply

Your email address will not be published. Required fields are marked *